John Quiggin argues that Charles Murray is another one of the American Enterprise Institute scholars who has materially contributed to its loss of scholarly reputation:
John Quiggin: The Flynn effect and the Bell Curve: In my last post in the AEI, I lumped Charles Murray, in with James Glassman, Karl Zinsmeister and Lynne Cheney as someone who had contributed to the loss of the AEI's reputation for scholarship (John Lott, the main subject of the post is in a class of his own).... I thought that this would be a good time to set out my views on Murray, and more particularly on The Bell Curve with Richard Herrnstein.... The Bell Curve got a pretty thorough hammering on statistical grounds when it came out (this Heckman review is one of the more favorable but is still pretty damning):
I would not have said that the Heckman review was "pretty damning." Any review that says that a book fails for four main reasons is more critical than that. I would have said that the impeccably right-wing Jim Heckman flayed Murry and Herrnstein alive and hung their skins on his office door:
Posted by DeLong at December 11, 2003 02:47 PM | TrackBackJames Heckman in Reason: ...The book fails for four main reasons. First, too much space is devoted to discussions of intrinsically irrelevant issues. Nothing central to the case for recognizing diversity in human abilities hinges on the issue of whether there is one "true ability" or whether there are multiple abilities--as common sense, much psychometric research, and the authors' own evidence indicate is the actual state of affairs. Despite this evidence, Murray and Herrnstein devote many pages to justifying a one-ability, or "g," model of human intelligence....
The second, more fundamental, reason why this book fails to provide an effective challenge to contemporary egalitarian social policy... [is that] the authors choose an empirical approach... [and] fail to develop the empirical case in a satisfactory or coherent manner.... [T]he authors do not discuss the costs and benefits of various interventions.... The authors seek to short-circuit all of the hard work required to make credible cost-benefit calculations by claiming that there is a genetic basis for skill differences. But estimates of a genetic component of skills are irrelevant to the requisite cost-benefit analysis unless it can be established that all differences are genetic. No one, including the authors, claims that this is so.... The Bell Curve fails to present the hard information required to settle these matters on the factual grounds chosen by the authors. This point is particularly telling for their assessment of education. The authors offer an inconsistent treatment of education throughout the book.... They acknowledge--and then go on to forget--that the relationship between education and ability is far from exact... throughout much of the book, they equate ability and education and implicitly assume that the economic returns to ability drive the economic returns to education....
[T]his implicit assumption is false. Their own evidence (buried in Appendix 6), as well as a vast literature in empirical social science, clearly indicates that controlling for ability lowers but does not eliminate the return to schooling measured in terms of earnings.... Ability and education are not the same thing, and both have economic rewards... Their implicit claim that ability drives the economic return to education, and the recent increase in the economic return to education, fails to pass empirical muster.
The third source of The Bell Curve's failure lies in the details of its analysis of the impact of ability on measured outcomes such as earnings. In their empirical research, the authors examine how well one measure of ability explains a variety of economic and social behaviors. They pit their ability measure against a measure of the socioeconomic status of persons when they were children. The authors intend this contrast to reveal the relative importance of "genes" and "environment" in accounting for behavior. Outcomes are much more sensitive to their measure of ability than to their measure of socioeconomic status. Large changes in the socioeconomic variables have weak effects on the outcome measures, while small changes in ability have large effects on the same outcome measures. This sort of empirical exercise prospers--or founders--on the details. The credibility of any empirical study depends on the care taken by the analyst in defining and measuring concepts, and in interpreting conclusions drawn from the data. It is at this point that the book becomes a policy polemic rather than a scholarly study of human differences....
Finally, the book fails due to a lack of coherence. The argument does not cumulate in a convincing way. Too many seams are visible...
I heard an American Enterprise Institute scholar (did not catch the name, sorry) say on "To The Point" (a PRI radio show) "87 billion dollars is not really alot of money."
I make donuts for a living. To net 87 billion dollars, I would have to make, oh, ~261 billion donuts. The AEI should learn more about American enterprise.
The whole matter of inheritance of intelligence, going back to Cyril Burt, has been dogged by incompetence-- and, in some cases, fraud. The plausible conclusion is that research in the subject is motivated principally by ideology rather than by disinterested scientific curiosity. Which is too bad, because it's obviously an important subject.
Posted by: Matt on December 12, 2003 06:06 AM"Nothing central to the case for recognizing diversity in human abilities hinges on the issue of whether there is one "true ability" or whether there are multiple abilities--as common sense, much psychometric research, and the authors' own evidence indicate is the actual state of affairs."
Indeed: take a n-dimensional vector, chances any group can't be defined as "superior" along some linear combinations of some of its components are slim... In other words, tell me which group you want to give some priviledges to, and I'll design a "resonable" IQ test that "shows" they are the smartest.
I am glad to see that Jim Heckman is as intellectually honest, even though right-wing, as he appeared to me when I had the chance to meet him a few years ago. This proves that, in some logical sense at least, there is an intersection between set "honesty" and set "right-wing". You would be tempted to think otherwise these days... [And fuzzy logics might lead you to a different conclusion ;-)]
Posted by: Jean-Philippe Stijns on December 12, 2003 06:35 AMI want to ask you again, Brad: have you ever crossed the campus to talk to your colleage Arthur Jensen about this? He might be able to clear up some of your confusion...
Posted by: jimbo on December 12, 2003 06:53 AMVarious right-wing IQ researchers post to the evolutionary psychology mailing list, which I read (including Charles Murray and Phillipe Rushton). Having seen them facing actual back-and-forth email debates over their ideas, I can safely say that they are complete fools. Every single one of them has the characteristic where any argument against their position, no matter how well posed, goes straight in one ear and out the other. They don't address criticisms of their positions, they dismiss them out of hand.
I read Flynn's famous argument against heritable racial IQ differences, and his model really devastates "Bell Curve" style models. His model argues for no significant differences, and explains all the data their model does plus one that it doesn't - the famous "Flynn Effect" where IQs have raised drastically in western countries since IQ measurement began (something like 15 points per generation). The Flynn Effect is completely incompatible with models that argue that straight correlations of current IQ with heredity imply that heredity has the overwhelming influence. Flynn's own model explains the effect and makes other very interesting predictions (that environmental influence on IQ is more temporary than permanent, but large, and that in many ways the environment drastically amplifies IQ differences by streaming high-IQ people into environments more likely to raise IQ).
When someone asks them about Flynn's model a common response is to simply dismiss it as some sort of semantic game that doesn't say anything differently than theirs does. What does it matter if heredity has a large influence on IQ, or indirectly has a large influence because the environment consistently amplifies IQ differences caused by heredity? Hey, it's all heredity underneath. (Except that Flynn argues powerfully that it's NOT heredity underlying intergroup IQ differences such as racial differences, and his model demolishes the political arguments that Murray and his fellow researchers like to base on their IQ research).
I have yet to see Murray or anyone adequately argue against the Flynn effect or their model. Anyone who can use basketball as an analogy for other types of learning earns my vote.
What I don't understand is how Bell Curve got to be a best seller and why so many in the media seemed to accept it uncritically?
Posted by: bakho on December 12, 2003 09:09 AMBecause people in the media are by and large innumerate. Even if they're well-meaning, they have a hard time sorting through mathematical-type arguments and are easily intimidated or frustrated as a result.
For instance, how many people in the media think that the US is a net EXPORTER of capital, for instance, while simultaneously complaining about the current account? We know that this is an impossible combination but I bet that not one person in 100 on the street, or in a newsroom, understands this.
Posted by: Chris on December 12, 2003 10:09 AM"What I don't understand is how Bell Curve got to be a best seller and why so many in the media seemed to accept it uncritically?"
Because it says what a lot of people, even in the media, would like to hear? "Blacks are stupid, poor things, and it isn't MY fault that they can't get ahead. Racism, like, died decades ago, dude!" (Here I'm mocking one sheltered ignoramus I had the misfortune of rooming with in my freshman year.)
Posted by: Abiola Lapite on December 12, 2003 11:54 AMCould some of the posters recommend some intro reading on this subject? I would be very interested. I didn't see a reference for the Flynn stuff.
Also, Arther S. Goldberger also has a very good critique of The Bell Curve -though many of the statistical mistakes in that book would be apparent to any one who paid attention in an intermediate undergrad class covering ANOVA and limited dependent variables regression techniques. The review used to be accessable from his home page.
In grad school I had a very good friend from Africa, who was a TA in geography. It amazed me how much math he could do in his head, and asked him where he got the skill. He said that pencils and paper were very hard to come by in country villages, and very expensive when found. So if you couldn't do math in your head, you would get taken to the cleaners whenever you went to market. He said that ordinary village girls could work everyday math problems much better than his undergrad US students with all of their pencil papers and calculators. I remember coming across him outside on the campus one day, trying to get some of his undergrad students to solve for the height of a flagpole. The students were completely befuddled. I joined in and tried to help him with the project. After awhile he got this look of simmering frustration on his face and I asked him what was wrong. He said that when he was their age, he could solve the problems in his head straight from a trig table. And he added "If this were in Africa, I could beat them for this." He was kidding, but that is *was* frustrating it was to teach the fancy US colleget students. Most of them were hopeless.
The issue of multiple abilities is fascinating. I remember reading about IQ studies that looked at different components of math reasoning across cultures. Africans consistently scored higher than Europeans and Americans on algebraic reasoning and numerical skill, while the latter two groups consistently scoured higher on geometric and spatial reasoning. And the distribution of specific skills in the populations were quite non-normally distributed. And what was interesting was that fundametal logic behind the algebraic problems and geometric ones were the same. So how does that work into the nature versus nurture debate, and the idea of a general 'g' versus the existence of independent and specific skills?
I will try to get the reference for the study I am quoting, but it has been years since I read about it. Maybe this weekend, when I can get to a good university library, if anyone is still posting here.
Here is the URL for Arther Goldberger and Charles Manski's review of The Bell Curve:
http://www.ssc.wisc.edu/irp/featured/bellcurv.htm
Posted by: jml on December 12, 2003 12:51 PMMurray and Herrnstein's book was valuable because it explicitly showed how cognitive ability predicts social outcomes. This may not sound like such a big idea now, but very few people in sociology or economics had bothered to look at it before TBC (and a lot of scholars who should, still don't).
The controversy comes in when trying to change cognitive ability. It's all fine and good to talk about the Flynn effect, but almost every deliberate intervention (e.g. Head Start) that has been tried, has had no lasting effect. Even the effects of adoption seem to fade after adolescence - a point that Flynn himself emphasizes. Until someone figures out a way to permanently increase cognitive ability, the right-wing IQ people are not unjustified in their skepticism.
Posted by: tc on December 12, 2003 01:49 PMI think tc might want to take a look at Goldberger and Manski's critique (see reference above) of the methods used to estimate the extent to which cognitive abilities determined social outcomes. The methods were very doubtful (uncharitably put, mistaken) and the results confounded any true effect with the initial distribution of abilities and socio-economic status existing in the sample population. So I don't think the Bell Curve is the place to look for reliable information on this topic.
Posted by: jml on December 12, 2003 02:46 PM"Until someone figures out a way to permanently increase cognitive ability, the right-wing IQ people are not unjustified in their skepticism"
Um, except that we HAVE found ways to permanently increase cognitive ability. That's what the Flynn effect IS - a continuous and substantial growth in the average IQ of the population over time. The magnitude is huge - something like 15 points per generation if I recall correctly. The "problem" is that, apparently, the stimulus has to be pretty much permanent or at least very prolonged - as education and knowledge-oriented skills become more important for people of all ages, IQ shows substantial increases. But there's no magic intervention where you can get 'em young, teach 'em for a couple years, and produce a dramatic lifelong benefit. Ultimately, once a poor kid in the ghetto gets finished with Head Start, they're still a poor kid in the ghetto, with much less prospect of continued intellectual stimulation than a middle class kid in the suburbs.
To give a direct example, there's no reason to expect blacks in the US to have the same average IQ as whites as long as they continue to have poorer economic and educational opportunities from the minute they're born (due to poor socioeconomic background, poor neighborhoods, inequitable allocation of school resources, and various forms of racism). Their low IQ is the product of living their whole lives in an inegalitarian society, not a product of living a couple of "critical years" of infancy in an inegalitarian society.
Flynn's model provides potent justification for egalitarianism - just not for US-style "bandaid on a gaping head wound" social policies.
BTW Flynn's paper can be found here:
http://www.apa.org/journals/rev/rev1082346.html
"I think tc might want to take a look at Goldberger and Manski's critique (see reference above) of the methods used to estimate the extent to which cognitive abilities determined social outcomes."
A lot more research has been done since TBC and its reviews were published. In particular, cognitive ability has been shown to predict social outcomes among siblings raised in the same family, and that's just about the ultimate way to control for family background. See Korenman & Winship (2000), "A reanalysis of The Bell Curve: Intelligence, family background, and schooling" in _Meritocracy and Economic Inequality_, eds Arrow, Bowles, and Durlauf.
"Um, except that we HAVE found ways to permanently increase cognitive ability. That's what the Flynn effect IS - a continuous and substantial growth in the average IQ of the population over time."
The whole population, sure. But what about people or subgroups within a population? As Flynn points out in his paper, even adoption doesn't seem to have any long-term effects. If there is a way to raise cognitive ability within a society, it's going to have to be something even better than adoption.
Posted by: tc on December 12, 2003 11:35 PMThank you, tc, for the references on subsequent research. I still won't trust the Bell Curve itself, though.
Posted by: jml on December 13, 2003 02:36 PM"Um, except that we HAVE found ways to permanently increase cognitive ability. That's what the Flynn effect IS - a continuous and substantial growth in the average IQ of the population over time."
The whole population, sure. But what about people or subgroups within a population? As Flynn points out in his paper, even adoption doesn't seem to have any long-term effects. If there is a way to raise cognitive ability within a society, it's going to have to be something even better than adoption."
As I said, decrease inequity, racism, and so forth. IQ of a population subgroup will be less than that of the majority as long as the subgroup is disadvantaged in the classic ways - education, employment opportunities, and so on. So for example the low IQ of blacks is a result of their shitty social situation, not a cause. I don't think anyone on either side of the debate really considers IQ test scores to be the goal of social policy. The concern was whether IQ differences should be viewed primarily as a cause of, or a result of, intergroup inequity. The evidence seems quite predominantly on the side of "result of". The lesson to be drawn from this is "don't worry about IQ, worry about education, employment opportunities, discrimination, and so on". Certainly, don't look for interventions that will magically raise the IQ of a class of people independent of the rest of their position in society.