Stephen J. Dubner: My Part in his Upfall

by Henry on March 21, 2012

So it appears that Stephen J. Dubner, co-author of _Freakonomics_ is “upset at various critics”: He is deeply unhappy with Andrew Gelman and Kaiser Fung, for having written what appeared to me to be a skeptical but intellectually generous take on the Freakonomics project. He is angry at Ezra Klein, whom he describes as someone who is ‘in the business of attacking at any cost’ on the basis of a tweet that Dubner presents in a “rather misleading fashion”: And he believes that ‘a man named Chris Blattman’ (great title for a band btw), was insufficiently abject in his apologies for a post “in which he suggested”: that _Freakonomics_ did not provide sufficient credit to other bloggers. Dubner is entirely right when he suggests that apologies should not be self-serving. So I hope that my own apology – long overdue – is not misinterpreted as same. I’d hereby like to sincerely apologize for having done my little bit to make Stephen Dubner and the whole _Freakonomics_ phenomenon what they are today.

Long-time readers will be familiar with the _Crooked Timber_ seminar that we did many years ago on the original _Freakonomics_ book. I can’t say what exact role it played in helping the book becoming the mass cultural phenomenon that it did, but the publicists seem to think that it played a significant role in generating publicity. Part of this was likely novelty – no-one had done anything quite like this before, so that lots of other bloggers linked to it. The revised and updated paperback edition of the book described the seminar as having provided the most astute analysis to date of the book’s arguments.

Doing this seminar was, I’m afraid, my initiative. I could try to defend myself. I (and others) were more interested in Levitt’s original academic work than the popularization. We sort of said this _sotto voce_ in the seminar, but only _sotto voce._ Nor has _Freakonomics_ been entirely bad. It’s gotten e.g. Justin Wolfers, who is excellent value for money, out into broader public circulation.

But even if it seemed a good idea at the time, I should have known better. Yes – Levitt is an interesting and original economist, but the glib contrarianism and breezy confidence that silly econometric results would tell us something valuable about the world were baked into the cake from the beginning of the _Freakonomics_ project, and perhaps before. D-squared’s perhaps never-to-be-published CT summation of his various posts on Freakonomics makes that clear. It’s a bit like one of those high-end fashion marques that begins with haute couture, and ends up over-extending its brand by using it on everything from cheap plastic novelties to toilet paper.

Both the blog and the second book were pretty dreadful. John has written about the contrarianism of the book, while as Andrew Gelman has hinted, he could have been a lot nastier had he wanted to be, pointing e.g. to the blog’s highlighting of results suggesting that ESP works, that the economy wasn’t actually all that bad in October 2008 etc. Nearly every time that I’ve seen _Freakonomics_ mentioned in the last several years, I’ve felt guilty and embarrassed that I had something to do with its rise to prominence. Very likely, it would have become prominent anyway (it had a very well organized PR campaign). But perhaps, given the chanciness of social contagion etc, it would not.

In any event, there really aren’t any excuses. I’m genuinely sorry for whatever push I gave to help start the _Freakonomics_ snowball rolling down the hill. There’s not much I can do about it now, but there you go.



paul 03.21.12 at 4:22 pm

So when can we see D-squared’s summary judgment? Can you really withhold it now? Or has he done it at his own place?


Daniel 03.21.12 at 4:46 pm

I did, in a series of posts over time. I will try to get my act together but certainly not before the weekend.


adam.smith (was Sebastian(1)) 03.21.12 at 5:05 pm

I’m not sure if “sorry” is going to cut it here…
More seriously, I wonder what did go wrong. Is Andrew right, and the problem is really that the Freakonomics project went downhill, or is Henry right and the problem was there from the get go and he (and many others, me included) did not see it. I’m tending towards the latter view, especially after reading John DiNardo’s terrific reviews of the 1st book.
in that case, of course, the question is – “why were we so blind?”. My guess would be that we think causal inference is cool and that we were excited about a book that made causal inference look cool – and were willing to overlook (or unwilling to see) the Gary Becker-y bullshit as well as Dubner’s grandiosity.


Steve LaBonne 03.21.12 at 5:21 pm

That’s what happens when you fail to apply the following heuristic: stop paying attention as soon as you see the words “University of Chicago economist”.


Daniel 03.21.12 at 5:28 pm

Is Andrew right, and the problem is really that the Freakonomics project went downhill, or is Henry right and the problem was there from the get go and he (and many others, me included) did not see it.

with nearly anything that is both ubiquitous and appalling, be it U2, Communism or “Sex and the City”, it is almost always the case that a) it really did have something good about it at the beginning, and b) that the seeds of what it later turned into were always there right at the start if you knew what to look for with the benefit of hindsight. So these aren’t mutually exclusive.


john b 03.21.12 at 5:29 pm

There is a chronology point as well: at the time of the first book, only the most contrarian of folks thought the situation was an insane unsustainable boom, and therefore everyone’s underlying assumptions weren’t discredited nonsense.


hartal 03.21.12 at 5:52 pm

Does Levitt and Donohue’s study of the impact of legalized abortion on crime rates count as part of Freakanomics? I never read the book but do remember reading that paper more than ten years ago. I remember being puzzled by implicit counterfactual in the paper: had abortion not been legalized (and, I presume, effective access to abortion increased) in certain states, crime rates would have been higher in those states. I also don’t know how well the statistical analysis has stood up.


Daniel 03.21.12 at 6:14 pm

I also don’t know how well the statistical analysis has stood up.

It had a massive embarrassing blow-up, complete with programming and data errors. Levitt had a response which I thought was quite weak, but I think the analysis survives weakened. The basic problem in my mind though was that in a period when you have a massive exogenous shock like the crack wars, it is not necessarily the case that any manipulation of the data is going to get it to cough up answers – it’s like using sophisticated analytical methods to try and take the eggs out of a cake.


bay of arizona 03.21.12 at 6:16 pm

had abortion not been legalized (and, I presume, effective access to abortion increased) in certain states, crime rates would have been higher in those states. I also don’t know how well the statistical analysis has stood up.

Abortion was legalized in NY 10 years before Roe, and there was never any proof that there was a 10 year headstart in crime rates there, or a 5 year head start in Calif. I believe that they admitted in the NYT blogpost that the correlation between crime and lead in gasoline or paint was stronger.


Barry 03.21.12 at 6:22 pm

Daniel, I join in the call – post those critiques!


adam.smith (was Sebastian(1)) 03.21.12 at 7:19 pm

@bay of arizona:
Abortion was legalized in NY 10 years before Roe, and there was never any proof that there was a 10 year headstart in crime rates there, or a 5 year head start in Calif.
with all due critique to Freakonomics – state-by-state variation and earlier dips in crime rates in states that legalized abortion earlier is one of their core identification strategies. See section B starting on p. 395 of their paper:
This strategy is also discussed at length in the book. Whether the finding holds or not, it’s one of the better sections of the book, a really neat example of how to explain what an “identification strategy” is.


adam.smith (was Sebastian(1)) 03.21.12 at 7:20 pm

grmpf – the first two lines – from “Abortion” to “Calif” are supposed to be a quotation.


Phil 03.21.12 at 8:42 pm

I can’t honestly say I’ve read the whole of that Dubner post – I started to glaze over when I got to the “why didn’t he talk about how brave and honest we were that other time?” part, and by the time I got to “and he talks about his friends too, just like we do, except that we don’t, so there” I was skimming pretty quickly. (Edwards seems to feel no shame in admitting that he didn’t even read to the end…) But my, what a petulant wanker. The Blattman Correspondence is one for the ages. (“I realised that my previous email had made me sound like a bit of a petulant wanker, but I was still cross and it still wasn’t fair, so I wrote back…”)


Tedra Osell 03.21.12 at 9:11 pm

For what it’s worth, I’m pretty sure that the reason I’ve never read the book or the blog and generally had the impression that they were pretty poor is because of something I read on CT back in the day.


Neville Morley 03.21.12 at 9:49 pm

…it’s like using sophisticated analytical methods to try and take the eggs out of a cake.

Yup, if that’s in any way typical, another vote for a consolidated summary, not least for those of us who hadn’t found CT at the time. And can I use that line, with due acknowledgement, to characterise some of the weird things people try to do with statistics in pre-modern economic history?


bianca steele 03.21.12 at 10:04 pm

That diagram is pretty stunning (the one at the Am. Sci. link).


John Quiggin 03.21.12 at 10:14 pm

I’m going to defend my part in the upfall of the original book.

First, I can scarcely regret helping to promote the production of econ books with garish covers and pop culture titles. And I think that the net impact of this class of books has probably been positive, at least as compared to a lot of the pop economics that we get.

Second, my contrarianism sensors weren’t quite as developed then as they are now, but I don’t think Freakonomics was particularly contrarian. The abortion stuff was controversial, but I think it’s fair to defend it (as I did at the time) as following the data where it leads (or it least where it leads your econometrics package). The rest of it was clever data analysis leading (in most cases) to intuitively plausible conclusions (eg: given strong incentives to produce good test scores, some teachers will cheat). The work may have some problems, but the contrast between that material and Superfreakonomics, where Levitt uses three years data to proclaim that the climate has been cooling, is so great that it’s hard to believe that it’s the same person. (Not the only instance where success of this kind has been very bad for the writer’s academic standards).

On the other hand, I admit to being susceptible to the breezy confidence of the book, and to being inadequately critical. It was, as Henry says, one of the first events of this kind, and enthusiasm for the project carried over to enthusiasm for the book.


Gene O'Grady 03.21.12 at 10:37 pm

I haven’t read the book, but I did look through at Borders or some such place once, and found (I think it was part of some teacher bashing) the assertion that the method of agreement in error was (a) new and (b) a contribution of economics. In point of fact this method (associated, not completely accurately, with Karl Lachmann) was revolutionary in classical studies in the early 19th century, and apparently came there from Biblical criticism, and may have originated in medieval MSS studies. Interestingly, I bought an old copy of the Edinburgh Review which touched on this question in 1850.

So that was about as far as I got, since I have trouble tolerating the efforts of economists as intellectual imperialists.


hartal 03.21.12 at 10:53 pm

I don’t know, John; the abortion stuff seems hardly contrarian. The basic idea is that the unwanted children born to families who are not in a position to nurture them are more likely to be convicted of a crime. But so many prejudices remain untouched by (or reinforce) this sensationalist analysis.

First, it takes the crime statistics at face value. For example, they don’t show that these unwanted poor children are responsible for the real rise in white collar crime that often escapes prosecution. Perhaps rates of non-violent drug offenses were higher in pro-life states. So they show some relation between abortion and reported crime?

Second, bad or unsupportive mothers are implicitly blamed for kids turning out bad. But what if those states that were more pro life were also more punitive toward the poor. Perhaps those kids born in pro-life states would have had more opportunities in pro-choice states and met less prejudice? At any rate, there is no good reason to believe the counterfactual: had these children been born a disproportionate number of them would have committed crimes. The assumption here is some of kind of genetic/early nurture determinism.

So I see a rather conventional analysis in terms of the sociology of crime, patriarchal assumptions and determinist ideology.

They are not C. Wright Mills or Pierre Bourdieu.


Phil 03.21.12 at 11:17 pm

I see a rather conventional analysis in terms of the sociology of crime, patriarchal assumptions and determinist ideology.

Which I think is a large part of Freakonomics’ weird combination of apparent contrarianism and mainstream success. That sleeve-tugging “maybe the nerd has something to say” shtick – “but what if – no wait, hear me out – what if *when we look at the data*…” – was tied to ideas which were only controversial because we’d grown up believing them and then rejected them. The illicit thrill of Freakonomics is the thrill of thinking “what if my grandparents, the cranky old man next door and James Q. Wilson were right all along?”


tomslee 03.21.12 at 11:50 pm

hartal – I think a lot of the appeal of the book is that it argues against positions such as yours. If I put on a Freakonomics-as-best-I-remember-it hat I’d argue that:

The basic idea is that the unwanted children born to families who are not in a position to nurture them are more likely to be convicted of a crime.

Not at all. The basic idea is to look at data without preconceptions and see where it leads us, rather than cutting off explanations because we didn’t like them.

bad or unsupportive mothers are implicitly blamed for kids turning out bad. Ditto.

The assumption here is some of kind of genetic/early nurture determinism. Ditto. No assumptions of any kind.

I see a rather conventional analysis in terms of the sociology of crime, patriarchal assumptions and determinist ideology. Not at all. If the data lead us to conclusions that happen to be conventional, so be it. If not, so be it.

And I have to say, I find something appealing about that approach. If it went over the top and down hill, not to say round the bend, later on well that’s too bad.


tomslee 03.21.12 at 11:54 pm

That said, the “case of the missing girls” highlighted in the Gelman & Fung piece was a pretty bad case of hubrisonomics.


hartal 03.22.12 at 12:04 am

Facts do not speak. They accept conventional measures of crime; those are produced facts. Unemployment does not simply exist; there are questions about how to measure and count it. A simplistic fact/value distinction is not tenable. I haven’t read Putnam about this, but have learned from Michael Root and Harold Kinkaid, ed. on this question.
To the extent that freakanomics reinforces a commitment to an untenable positivism, it stands in the way of, and is not an expression, of critical thinking.

I should not trust my memory, but if I remember correctly the freaking economists do themselves say explicitly that legalized abortion did in some places and would have in other places reduced (conventionally measured) crime because poor unwanted children do commit crimes at a disproportionate rate. That strikes me a simplistic and, alas, sensationalist claim.

At any rate, I also raised a simple question: were pro-life states more likely to have more punitive policies in regards to poverty and drugs?


hartal 03.22.12 at 12:16 am

From my old notes, I find that this is in the prepublication paper

“A number of studies have looked at cases of women, living in jurisdications in
which govt approval to have an abortion was required, who sought to have an abortion, but were denied to the right to do so. Dagg (1991) reports that these women overwhelmingly kept their babies, rather than giving them up for adoption, but that they often resent the unwanted children and, perhaps because of feelings of depression and anger, were far less likely than other mothers to nurture, hold, and breastfeed their children. In an array of studies in Eastern Europe
and Scandinavia, Dagg found that the children who were born because their
mothers were denied an abortion were substantially more likely to be involved in crime and have poorer life prospects, even when controlling for the income, age, education and health of the mother. This literature provides strong evidence that unwanted children are likely to be disproportionately involved in criminal activity, which may be the causal pathway from greater availability of abortion to lower rates of crime.”

Here is an unmediated link between neglectful mothering and crime as conventionally understood. Rather simplistic.


Manoel Galdino 03.22.12 at 12:44 am

There is no such a think as let the data speak. First, there is your model (linear regression? Clustered standard errors? etc.). Second, you do need to think of a causal effect, i.e. some mechanism to explain an association between two variables, even with clever identification strategies (if not, you can’t generalize, for instance). Last, but not least, you have the questions you looked upon in the first place and that have to fit with other theories out there (about how mothers impact the prospects of kids, for example).

All in all, I don’t buy the kind of argument about the data. However, I do buy the argument that we should adjust our theories based on the data (and all three aspects above).


Helen 03.22.12 at 12:47 am

More seriously, I wonder what did go wrong. Is Andrew right, and the problem is really that the Freakonomics project went downhill,

I’m only a layperson, but could I suggest that the Freakonomics project was inherently unsustainable. The first Freakonomics book was a “best of” from people who seemed to have, in perfectly good faith, come up with some explanations of common phenomena which were exciting because they were unexpected. Once they have that book advance for the followup, they have to come up with these discoveries. It follows that they’ll be more likely to duck and weave a little more in their approach to the topics – e.g. the “global cooling” schtick.

Like the Second Album syndrome writ large.


harry b 03.22.12 at 1:37 am

Helen’s right. Its surprising, then, that there isn’t a chapter in Superfreakonomics about the phenomenon she describes. You’d think such smart contrarians would have anticipated and studied the predicament they entered, really.


Harold 03.22.12 at 2:10 am

I always felt that the unspoken and very mischievous and racist implication of Freakonomics was that crime fell because those aborted were black, and that blacks commit crimes not because their mothers are neglectful (much less for economic or social reasons) but because they are black and therefore essentially prone to criminality. This supports in a back-handed and unflattering way the rumor one used to hear put about that abortion was a kind of genocidal plot by hypocritical white liberals who in reality secretly racist. Dubner wrote a very interesting and touching autobiography, granted. But Freakonomics is the vilest sort of garbage, IMO, on a level with the Protocols of the Elders of Zion.


hartal 03.22.12 at 2:52 am

Too quick, Harold. Levitt cites studies about neglectful mothers in Scandinavia and Eastern Europe.

Of course in America black mothers are thought to be generally neglectful mothers, with high genetic chance of giving birth to demon spawn.

We have to explain the phenomenon of Freakanomics, and a crucial factor here, noted by both you and Phil at 20, is that readers found their prejudices confirmed in the apparently hard-nosed analysis, and Levitt did nothing to stand in the way, taking cover as a value-free, positivist social scientist who was simply letting the data speak.

Another typical Faustian bargain with race, just beneath the surface here of another American success story.


Harold 03.22.12 at 4:32 am

Well, he only cited Scandinavian and Eastern European neglectful mothers as a shield, then.


piglet 03.22.12 at 5:48 am

Just read the take-down by DiNardo (thanks to 3 for the link) and must say that it is devastating. If you haven’t read it, is a short and rather easy read. Although written far more charitably than deserved, it shows that the underlying science is and was always sloppy. The other paper,, is longer and much more complex but absolutely worth reading.


hartal 03.22.12 at 6:31 am

I don’t think DiNardo is getting at what remains questionable about the legalized abortion and crime piece. In fact DiNardo admits that the corrected story about the effects of the Romanian ban on legalized abortion is not necessarily inconsistent with Levitt’s story about the US.

What some of us are saying here is that without the legalized abortion story and Levitt’s allowing people to read their prejudice into it which in turn raised concerns about negative eugenics, he would never have come center stage. And there would have been no Freakanomics phenomenon. His American success story is rooted, as American success so often has been, in race.

Why didn’t Devah Pager become a phenomenon? She has only showed that black job candidates are as likely to be called back as white candidates with resumes that are identical but for the addition of a felony criminal record.

That is a more interesting finding to me.

The legalized abortion story is really the simplistic and sensationalist one that “crime” is caused by the demon spawn of poor neglectful mothers. It’s a simple story about crime that appeals to white middle class prejudice. It gives white middle class people evidence for their smug Malthusian beliefs.

No, children born to mothers who are not ready for them do not disproportionately become criminals without a lot of other factors in play, e.g. lack of decent child care, bad schools, prejudice in the labor market, poor prospects for adoption for poor colored children, etc.

But you want to reduce crime? Well a most effective way is just to legalize abortion, Levitt’s research seems to show. Ignore the other stuff, just like Levitt and Donohue did.

The pro-life states were most probably also more punitive in their policies towards poverty and drugs; they probably allowed more discrimination in the labor market. They probably did a lot of things to make life more difficult for poor kids than the pro-choice states. After all, the pro-life states are disproportionately Southern, which was suffering a white backlash against the Civil Rights movement in the post Roe v. Wade years.

Is it so simple that crime did not fall in pro life states because poor mothers in particular could not abort their unwanted children?


MS 03.22.12 at 7:25 am

There’s something strangely gratifying about this post simply because Freakanomics annoyed me so much when I read it (very cursorily, while I was house sitting for someone). Since I am not an economist and the argument is data-driven, I had absolutely no way to disprove the data and had this awful feeling that I should accept the conclusions. It’s been so long I don’t remember exactly what it was that bugged me the most-something about prostitutes, I think.

Malcolm Gladwell also drives me crazy, if you want to go after him next.

On the one hand, I think it’s great when people discover new fields by populizers. It seems a shame when people refuse to write an accessible or exciting book for non-specialists because they open themselves up to ridicule. On the other hand, as a non-specialist, you are very captive to the argument. So it’s a very bad thing for people to do a sloppy job. As much as I disliked the approach, I didn’t get the sense the author didn’t respect his audience and was being sloppy for that reason. When I read about some of the mistakes he made though, I wonder.


Phil 03.22.12 at 11:25 am

To be clear, I don’t doubt that Levitt and the people he cited in the first book were following where the data led, setting aside their preconceptions and so on. But you don’t have to be Bruno Latour to think that the data will lead different people to different conclusions, and that what remains when you’ve set aside the preconceptions you’re aware of isn’t necessarily a blank slate. If you’re socialised into a culture of unreflective authoritarian individualism – people work hard and they do well, people go bad and they get punished – the “common sense” substrate of your beliefs, how you believe the world is, will always bear traces of that culture. You can set it aside if you really try, but it will be far easier to put aside preconceptions you learned at a later date – things like human equality, the importance of procedural justice, the complexity of social causation. I think that’s what happened with L&D, and that’s why the book was such a hit.


Zamfir 03.22.12 at 11:40 am

Nah, Gladwell is really different. He might be even more likely to tell a story and fudge the facts to fit, but he doesn’t have the authority of being a Real Scientist.


J. Otto Pohl 03.22.12 at 12:23 pm

I read the first Freakonomics in a book store in Alexandria, VA. It was during my three years of unemployment after finishing my PhD from SOAS so I had no money to buy books. But, I did have the time to read it. It amused me as kind of a series of intellectual parlor tricks. The authors were able to show counter intuitive results for a number of things. The piece I remember the most was why most drug dealers were poor. Their profession like acting and university teaching being based on a tournament model where a few became very rich. But, I don’t remember anything in the book that was more than an amusing “you would think, but you would be wrong.” It did not strike me as being a serious intellectual attempt to reshape popular thinking about economic matters.


tomslee 03.22.12 at 1:06 pm

I third Helen.

On the facts thing. Whenever I see “facts do not speak” (hartal @23) I’m seized with an impulse to ask “Is that a fact?” I do agree that many claims to just follow the data are clouded in preconceptions and ideology, and hartal’s example at #24 suggests the Freakonomists are indeed guilty of this. But to argue backwards from the conservativeness or conventionality of the conclusions to the tenability of the theory (emphasizing the frailty and unstubbornness of mere facts), as hartal does, seems to me a failsafe way of avoiding changing your mind, which is surely conservatism of the highest order.

Maybe I am just simplistic, to use a word hartal employs no less than four times. I guess I’m just not up to the kind of sophisticated postmodern contortionism that is required to think unconventionally.


tomslee 03.22.12 at 1:11 pm

Just to be clear, I plead guilty to contrarianism at 21, and hereby fully endorse Phil’s reading of the book.


bexley 03.22.12 at 1:36 pm

Helen @ 26

Further evidence for your theory is that the last chapter of the first book was the worst. It was something about trends in naming children but the impression I was left with was a chapter filled with long lists of first names. Suggests they were already running out of material back then.


hartal 03.22.12 at 2:04 pm

OK, tomlee, let’s start with the facts. We know what Levitt’s main explanation is for the fall in the crime rate beginning in 1991, but how does he explain its rise after the mid 60s to the 1980s.
Let’s just take Levitt’s facts.
Homicide rates were fairly stable at 4-5 per 100,000 from 1950 through the mid 1960s, at which point they started rising to a peak of 10.2 per 100,000 in 1980. From 1980-1991 the homicide rate fluctuated between 8-10 percent 100,000. After that, the homicide trend began a large, steady decline. Between 1991 and 2000, homicide rates fell from 9,.8 to 5.5 per 100,000, a drop of 44 percent.


hartal 03.22.12 at 2:23 pm

Check out the Journal of Economic Perspectives piece by Levitt in 2004. I can’t find any attempt to explain the rise in the homicide rate beginning in the mid 1960s before it began to fall in 1991. Perhaps I am missing it. If not, that would suggest that there is something quite strange about the analysis.


marc 03.22.12 at 3:01 pm

@J. Otto Pohl 03.22.12 at 12:23 pm

It did not strike me as being a serious intellectual attempt to reshape popular thinking about economic matters.

A common misconception: Freakonomics is not at all about economics – unless you understand economics as regression plus some reliance on a notion of ‘incentives’.
(Having said that- add more maths, some notion of maximisation and you arrive at what many economists do ;-))


hartal 03.22.12 at 5:02 pm

Again, looking at Levitt’s own data, one could easily guess that the homicide rate rose sharply after the mid-60s due to some combination of the complex effects of a war in which 50,000 Americans died killing one million Indo-Chinese, periods of very high unemployment (stagflation, 1980-2 recession), and the emergence of crack in the 80s. It seems hardly surprising that these causes would weaken in the 90s and or that the murder rate did not fall as sharply in those Southern states that were, among other things, more pro-life. Those states were probably doing or not doing lots of other things–not reducing lead exposure, implementing incarceration strategies that led to more homicide, criminalizing more behavior, cutting back more sharply on welfare and social support.
But Levitt had a simple story, the most effective and cost-effective way to reduce crime is to allow poor, would-be-negligent mothers to abort their demon spawn. The story confirmed the prejudices of the white middle class people looking for books at airports.
And we got the Freakonomics phenomenon, another sad chapter in American intellectual life


hartal 03.22.12 at 5:28 pm

In a blog defense, Levitt gives his strongest piece of evidence first “1) Five states legalized abortion three years before Roe v. Wade. Crime started falling three years earlier in these states, with property crime (done by younger people) falling before violent crime.”
Yet–as someone at some point noted– since women came from adjacent states to have abortions (say from Connecticut and New Jersey to New York), crime should have fallen earlier in them too. Yet Levitt is saying that this did not happen. This data no more confirms than falsifies Levitt’s hypothesis.


piglet 03.22.12 at 6:04 pm

DiNardo cites at least two (?) examples where Freakonomics misrepresented others’ work (one case was the Romania abortion ban, the other … ?). DiNardo is ostensibly charitable about it but for me, such misrepresentation alone disqualifies the authors. Some of the mistakes may be “mere” sloppiness but clearly there is also academic dishonesty.


Phil 03.22.12 at 7:38 pm

What gets me about the abortion/crime study is that the hypothesis doesn’t pass the smell test. It’s the kind of speculation that would make most social scientists think “what would the world have to be like in order for this to be true?”, “what policy implications would follow from treating this as useful?” and then think “might there be other hypotheses with less repugnant implications which the data would satisfy just as well?” Deliberately ignoring those alarm bells isn’t being brave and iconoclastic, it’s boldly going where half of American criminology has gone before – only going further. (The difference between Freakonomics and Broken Windows is much smaller than either side’s advocates make out; it’s the difference between “crime means that these people need to be controlled” and “crime means that these people need not to be born”.)


Harold 03.22.12 at 8:39 pm

Phil, exactly. Except that broken windows applies to everybody, not just “these people,” at least that’s how I understood it. Also, Piglet, Dubner misrepresented Stetson Kennedy in a gratuitously egregious and insulting way. If he engaged in all these misrepresentations, does this not raise the possibility (probability) that there were other misrepresentations, perhaps even in his striking autobiography.


Harold 03.22.12 at 9:10 pm

What about Dubner’s “contrarian” conclusion that terrorism by the Klu Klux Klan was really a good thing for the South because it maintained law and order and cut crime?


Phil 03.22.12 at 9:58 pm

broken windows applies to everybody, not just “these people,”

Don’t think so.

“Many citizens, of course, are primarily frightened by crime, especially crime involving a sudden, violent attack by a stranger. … But we tend to overlook another source of fear–the fear of being bothered by disorderly people. Not violent people, nor, necessarily, criminals, but disreputable or obstreperous or unpredictable people: panhandlers, drunks, addicts, rowdy teenagers, prostitutes, loiterers, the mentally disturbed.”

Solution: move ’em on, kick ’em out, keep ’em quiet.


anon 03.22.12 at 11:18 pm

I would be really interested to see the hits their site receives prior to and after the GFC. It seems to me – wholly unsubstantiated by data but if I had the data I’d like to test this – that Freakonomics was part and parcel of the moment of the great moderation hubris. One assumes that now the field remembers that it has better things to do than show Sumo wrestlers cheat (which was well known before they did the analysis and is not counter-intuitive at all) Freakonomics should wither and die.

Or one can hope, in any case.


Peter T 03.23.12 at 12:14 am

MS at 33 hits it. Books like this are interesting, but the lay reader is at the mercy of the experts. So the onus is on the experts to be honest and up-front. I glanced through Freakonomics and found major errors in the couple of fields I know something about. Which suggested that I could not trust it in fields I know nothing about – something confirmed by people who did know something in the other fields touched on. How many other books are there like this?

tomslee – facts may speak, but they do so in sentences, conversations, dialogues…Not in single words. Somebody who throws facts around without regard for context is unlikely to be making sense. As in “I’m freezing, so climate change is a myth”. No contortions required, just attention.


Harold 03.23.12 at 12:16 am

Well, that may be. I don’t think that Gladwell’s solution is what you propose. I don’t know where you live, Phil. I live in a neighborhood with its share of drunk and disorderly people (on a Saturday) and the others mentioned, and while I don’t fear them, I don’t necessarily like having them on my block and hearing them all night long, would you? I am happy when a bar is replaced by another sort of business, such as a hardware store.


tomslee 03.23.12 at 2:16 am

hartal: I really didn’t claim that the Freaks are right, just that they are not wrong because of the conventionality or otherwise of their conclusions. And yes, facts by themselves have at best a limited vocabulary.

On the abortion/crime issue – my recollection of the reaction at the time is that it was widely seen as generally a progressive conclusion in that it gave support to the pro-choice side of the abortion debate. Am I misremembering?


John Quiggin 03.23.12 at 5:02 am

@MS The prostitute stuff was in SuperFreakonomics, so your negative reaction was entirely justified.


Daniel 03.23.12 at 7:08 am

there was also the amazingly annoying and really quite culpably obviously counterfactual statement that “the typical prostitute earns more than the typical architect” in the first book, which was the first cockroach which originally tipped me off that something might be wrong …


Alex 03.23.12 at 9:27 am

, one could easily guess that the homicide rate rose sharply after the mid-60s due to some combination of the complex effects of a war in which 50,000 Americans died killing one million Indo-Chinese

Well, no, not unless you also had a copper-bottomed explanation as to why the (vastly bloodier) Second World War and the (comparably bloody) Korean War had no such effect.


Katherine 03.23.12 at 9:32 am

Freakonomics lost me when it asserted that if you were pro-choice, but anti-death penalty, you were holding those views entirely because of tribal loyalty, since those two positions were obviously mutually exclusive, logically speaking. But this book would be thinking in straight lines and looking at just the facts, ma’am.

I realised that anyone without the imagination to understand the concept of subjective, but nevertheless logically consistent, moral and ethical codes was going to be gobsmackingly arrogant about their own “objective” analysis. And entirely unaware of their own subjectivity. Straight line thinking was going to mean blinkers.


Harold 03.23.12 at 10:40 am

Well, I am glad they are collapsing of their own weight.


Phil 03.23.12 at 12:40 pm

while I don’t fear them, I don’t necessarily like having them on my block and hearing them all night long, would you?

I wouldn’t, & didn’t when I was in that position. But that’s very different from saying either what Kelling & Wilson actually said in their Broken Windows article or what they were subsequently interpreted as saying, which was (respectively) “the police should crack down on drunks, street people and unruly teenagers, just like they used to in Mythical Small-Town America, because it’ll make respectable people feel safer, which matters more than the actual crime rate” and “…because it’ll make the crime rate go down by some mechanism which nobody’s quite spelt out yet”. Unlike Kelling and Wilson, I don’t believe street drunkenness, homelessness or neighbourhood decline is a policing problem; I’m not at all convinced that crime is.

Tom – the trouble with seeing that argument as progressive is that the logical extension of it is that those women whose unwanted children would be particularly likely to offend should have particularly good access to abortion. This isn’t hypothetical – DiNardo quotes a notorious comment by William Bennett, which was conspicuously not disowned by Dubner. The reason it’s vulnerable to being extended in that way is that it rests on the understanding that a constant level of crime is caused by a constant supply of criminals, so the aims of criminal justice policy can be achieved by controlling or removing the criminals – pre-emptively in this case.


hartal 03.23.12 at 4:34 pm

Alex @ 56

I am not sure that WWI, WWI and the Korean War cannot be linked to spikes in the domestic homicide rate. The Vietnam War lasted, by some counts, ten years; its legitimacy and purpose seem to have been widely questioned even as it resulted in a catastrophic number of deaths, which in turn may have provided official sanction to pointless and illegitimate violence; and there was no GI program to assimilate the veterans.

It may not be surprising that the possible connection between war (especially widely perceived illegitimate war) and domestic homicide rates was not strong in the case of the short Baudrillardian Iraq War; the domestic homicide rate began falling in 1991.

I do not know the social science research on wars and domestic homicide rates. I am sure it exists, and it remains curious that Levitt provided no explanation for why the domestic homicide rate rose sharply between 1965-1990.

Perhaps the Vietnam War would not be part of the explanation, but then what would be?


hartal 03.23.12 at 4:37 pm

Phil at 59
That last sentence really does hit the nail on the head. Well put indeed.


Harold 03.23.12 at 6:17 pm

I did not actually read broken windows. From reports, I understood it to mean that if you allowed windows to stay broken then people would think nobody cared and would break more windows.

When I used to take my children to the park, I would bring a large broom and dust pan and sweep up the broken glass, hoping to set an example. I participated in volunteer weekend efforts to remove graffiti from the walls of the public library. I also went to a community meeting at the police station and objected when the police captain said it was useless to discourage people from building fires in the public park, because if you arrested people it was too much paperwork and they would just be let out anyway. He said that things were much better when he was young and the police used to just beat people over the head with a night stick and that was the end of it (this was uttered at a public meeting), and that when he was a teenager, he was beaten by police a couple of times and he was a better man for it. (I remember from my youth also that this was the way the cop on the corner enforced the law, after my stepfather was mugged while walking the dog in the 1950s. The police just beat up the suspect and that was it. They didn’t bother arresting him).

While I abhor the now-pervasive policy of stop-and-frisk on the basis of race (and concomitant false arrest and imprisonment), I don’t think “petty” crimes should be ignored because “it is too much trouble” to do things the constitutional way. But perhaps you are right, the fact that we have (or had, they have declined markedly) such crimes at all is a social, not a criminal law problem.


Danny Yee 03.24.12 at 10:59 am

My label for Freakonomics-style analysis is “drive-by regression”.


David in NY 03.26.12 at 9:23 pm

Re the rise in crime from mid-60’s to 1990 (comments 43, 56, 60). I, as a total amateur, have always favored the demographic correlation (just letting the facts speak for themselves here). During that period, due to the baby boom, there was a far greater proportion of the population between the ages of 15 and 30 than before or after, and a smaller proportion of adults to care for them. The bulk of crime is committed by young men in that age range. Voila, crime wave. (I think recent scholarship has disputed this, but don’t know why.)


hartal 03.27.12 at 5:44 am

Because the percentage of that population (the youth demographic) rose again in the 90s–Levitt calls this the echo of the baby boom–while the crime rate remained law.

Comments on this entry are closed.