February 22, 2003
Notes: What Thomas Sowell Thinks of The Bell Curve
Atrios laments: "Every time I refer disparagingly to [Herrnstein and Murray's] The Bell Curve some true believer expects me to write a 50,000 word critique of the book to justify my opinion of it. Frankly, it's as if every time I spoke disparagingly of The Protocols of the Elders of Zion someone expected me to write a 50,000 word critique of it."
You don't have to write a 50,000 word critique. All you have to do is point them to Thomas Sowell's American Spectator review of The Bell Curve. Sowell pulls his punches--no book that goes to the lengths The Bell Curve does to keep from considering education as an independent influence on people's life-paths can possibly be, as Sowell calls it, "very sober, very thorough, and very honest." But even Sowell's pulled punches are absolutely devastating. And no one can call Thomas Sowell a politically-correct left-wing hack:
Posted by DeLong at February 22, 2003 08:39 AM
Upstream: Issues: Bell Curve: Thomas Sowell
: Vol. 28, American Spectator, 02-01-1995, pp 32.
[Herrnstein and Murray] seem to conclude... that... biological inheritance of IQ... among
members of the general society may also explain IQ differences between different racial and ethnic groups.... Such a conclusion goes... much beyond what the facts will support....
[T]he greatest black-white differences are not on the questions which
presuppose middle-class vocabulary or experiences, but on abstract
questions such as spatial perceptual ability.... [Herrnstein and Murray's] conclusion that
this "phenomenon seems peculiarly concentrated in comparisons of
ethnic groups" is simply wrong. When European immigrant groups in the United States scored below
the national average on mental tests, they scored lowest on the
abstract parts of those tests. So did white mountaineer children in
the United States tested back in the early 1930s. So did canal boat
children in Britain, and so did rural British children compared to
their urban counterparts, at a time before Britain had any significant
non-white population. So did Gaelic-speaking children as compared to
English-speaking children in the Hebrides Islands. This is neither a
racial nor an ethnic peculiarity. It is a characteristic found among
low-scoring groups of European as well as African ancestry.
In short, groups outside the cultural mainstream of contemporary
Western society tend to do their worst on abstract questions, whatever
their race might be....
Perhaps the strongest evidence against a genetic basis for
intergroup differences in IQ is that the average level of mental test
performance has changed very significantly for whole populations over
time and, moreover, particular ethnic groups within the population
have changed their relative positions during a period when there was
very little intermarriage to change the genetic makeup of these
While The Bell Curve cites the work of James R. Flynn, who found
substantial increases in mental test performances from one generation
to the next in a number of countries around the world, the authors
seem not to acknowledge the devastating implications of that finding
for the genetic theory of intergroup differences, or for their own
reiteration of long-standing claims that the higher fertility of
low-IQ groups implies a declining national IQ level. This latter claim
is indeed logically consistent with the assumption that genetics is a
major factor in interracial differences in IQ scores. But ultimately
this too is an empirical issue--and empirical evidence has likewise
refuted the claim that IQ test performance would decline over time.
Even before Professor Flynn's studies, mental test results from
American soldiers tested in World War II showed that their
performances on these tests were higher than the performances of
American soldiers in World War I by the equivalent of about 12 IQ
points. Perhaps the most dramatic changes were those in the mental
test performances of Jews in the United States. The results of World
War I mental tests conducted among American soldiers born in
Russia--the great majority of whom were Jews--showed such low scores
as to cause Carl Brigham, creator of the Scholastic Aptitude Test, to
declare that these results "disprove the popular belief that the
Jew is highly intelligent." Within a decade, however, Jews in the
United States were scoring above the national average on mental tests,
and the data in The Bell Curve indicate that they are now far above
the national average in IQ.
Strangely, Herrnstein and Murray refer to "folklore" that
"Jews and other immigrant groups were thought to be below average
in intelligence. " It was neither folklore nor anything as
subjective as thoughts. It was based on hard data, as hard as any data
in The Bell Curve. These groups repeatedly tested below average on
the mental tests of the World War I era, both in the army and in
civilian life. For Jews, it is clear that later tests showed radically
different results--during an era when there was very little
intermarriage to change the genetic makeup of American Jews.
My own research of twenty years ago showed that the IQs of both
Italian-Americans and Polish-Americans also rose substantially over a
period of decades. Unfortunately, there are many statistical problems
with these particular data, growing out of the conditions under which
they were collected. However, while my data could never be used to
compare the IQs of Polish and Italian children, whose IQ scores came
from different schools, nevertheless the close similarity of their
general patterns of IQ scores rising over time seems
indicative--especially since it follows the rising patterns found
among Jews and among American soldiers in general between the two
world wars, as well as rising IQ scores in other countries around the
The implications of such rising patterns of mental test performance
is devastating to the central hypothesis of those who have long
expressed the same fear as Herrnstein and Murray, that the greater
fertility of low-IQ groups would lower the national (and
international) IQ over time. The logic of their argument seems so
clear and compelling that the opposite empirical result should be
considered a refutation of the assumptions behind that logic....
A man who scores 100 on an IQ test today is answering more questions
correctly than his grandfather with the same IQ answered
two-generations ago, then someone else who answers the same number of
questions correctly today as this man's grandfather answered two
generations ago may have an IQ of 85.
Herrnstein and Murray openly acknowledge such rises in IQ and
christen them "the Flynn effect," in honor of Professor
Flynn who discovered it. But they seem not to see how crucially it
undermines the case for a genetic explanation of interracial IQ
differences. They say:
The national averages have in fact changed by amounts that are
comparable to the fifteen or so IQ points separating blacks and whites
in America. To put it another way, on the average, whites today
differ from whites, say, two generations ago as much as whites today
differ from blacks today. Given their size and speed, the shifts in
time necessarily have been due more to changes in the environment than
to changes in the genes.
While this open presentation of evidence against the genetic basis
of interracial IQ differences is admirable, the failure to draw the
logical inference seems puzzling. Blacks today are just as racially
different from whites of two generations ago as they are from whites
today. Yet the data suggest that the number of questions that blacks
answer correctly on IQ tests today is very similar to the number
answered correctly by past generations of whites. If race A differs
from race B in IQ, and two generations of race A differ from each
other by the same amount, where is the logic in suggesting that the IQ
differences are even partly racial?
Herrnstein and Murray do not address this question, but instead
shift to a discussion of public policy:
Couldn't the mean of blacks move 15 points as well through
environmental changes? There seems no reason why not--but also no
reason to believe that white and Asian means can be made to stand
still while the Flynn effect works its magic.
But the issue is not solely one of either predicting or controlling
the future. It is a question of the validity of the conclusion that
differences between genetically different groups are due to those
genetic differences, whether in whole or in part. When any factor
differs as much from Al to A2 as it does from A2 to B2, why should one
conclude that this factor is due to the difference between A in
general and B in general?...
A remarkable phenomenon commented on in the Moynihan report of
thirty years ago goes unnoticed in The Bell Curve--the prevalence of
females among blacks who score high on mental tests. Others who have
done studies of high- IQ blacks have found several times as many
females as males above the 120 IQ level. Since black males and black
females have the same genetic inheritance, this substantial disparity
must have some other roots, especially since it is not found in
studies of high-IQ individuals in the general society, such as the
famous Terman studies, which followed high-IQ children into adulthood
and later life. If IQ differences of this magnitude can occur with no
genetic difference at all, then it is more than mere speculation to
say that some unusual environmental effects must be at work among
blacks. However, these environmental effects need not be limited to
blacks, for other low-IQ groups of European or other ancestries have
likewise tended to have females over-represented among their higher
scorers, even though the Terman studies of the general population
found no such patterns.
One possibility is that females are more resistant to bad
environmental conditions, as some other studies suggest. In any event,
large sexual disparities in high-IQ individuals where there are no
genetic or socioeconomic differences present a challenge to both the
Herrnstein- Murray thesis and most of their critics.
Black males and black females are not the only groups to have
significant IQ differences without any genetic differences. Identical
twins with significantly different birthweights also have IQ
differences, with the heavier twin averaging nearly 9 points higher IQ
than the lighter one. This effect is not found where the lighter
twin weighs at least six and a half pounds, suggesting that
deprivation of nutrition must reach some threshold level before it has
a permanent effect on the brain during its crucial early development.
Perhaps the most intellectually troubling aspect of The Bell Curve
is the authors' uncritical approach to statistical correlations. One
of the first things taught in introductory statistics is that
correlation is not causation. It is also one of the first things
forgotten, and one of the most widely ignored facts in public policy
research. The statistical term "multicollinearity," dealing
with spurious correlations, appears only once in this massive book.
Multicollinearity refers to the fact that many variables are highly
correlated with one another, so that it is very easy to believe that a
certain result comes from variable A, when in fact it is due to
variable Z, with which A happens to be correlated. In real life,
innumerable factors go together. An example I liked to use in class
when teaching economics involved a study showing that economists with
only a bachelor's degree had higher incomes than economists with a
master's degree and that these in turn had higher incomes than
economists with Ph.D.'s. The implication that more education in
economics leads to lower incomes would lead me to speculate as to how
much money it was costing a student just to be enrolled in my course.
In this case, when other variables were taken into account, these
spurious correlations disappeared. In many other cases, however,
variables such as cultural influences cannot even be quantified, much
less have their effects tested statistically....
Of course it's only one one of many factors, of course it may not be the dominant factor, but is it really so hard to believe that there are intelligence differences amoung the various breeds of people? If there can be siginifigant differences in other qualities, just compare the Pygmies and the Nuer peoples of Africa, why is it so taboo to consider intelligence variations?
"The Bell Curve" is rubbish.
>why is it so taboo to consider intelligence variations?
Because then unequal life outcomes become the province of the individual, and resistant to state-imposed equality imposition, aka quotas.
The Bell Curve book is a bloody shirt invoked by the left to justify screaming about racism. Rarely if ever is that book cited by a "right-winger" as proof or argument for anything.
People who have never read the book reflexively condemn it. They'll even admit it.
As to ethnic differences, this is my favorite humorous "Jewish" haiku:
Seven-foot Jews in
the NBA slam-dunk
My alarm clock rings.
Another GREAT scientist/writer who has reviewed the Bell Curve and other writings on race and genetics is the biologist Richard Lewontin. (Stephen Jay Gould said, "Lewontin is simply the smartest man I have ever met.") Much of his stuff appears in the New York Review of Books, and a few of his articles are available for free on its archieves at www.nybooks.com. If anyone has spare time in the few days and care tear themselves away from ongoing Iraq coverage, there is no better way to spend your time then reading his books, It Ain't Necessarily So: the Dream of the Human Genome and Other Illusions, and The Triple Helix: Gene, Organism, and Environment.
He's a genius, and it's a treat to read him. Hope some of you take him up.
Speaking of Stephen Jay Gould, let's not forget _The Mismeasure of Man_ which is essentially a book-length refutation of _The Bell Curve_.
The result Sowell quotes about the dramatically better performance of female blacks compared to male blacks is particularly remarkable.
If one presumes that it is almost entirely a set of environmental factors that accounts for this discrepancy, what it seems to demonstrate is that the differences in environment that make for differences in IQ scores are VERY subtle, and probably not at all the sorts of things that most people think of as crucial. For example, one expects that the basic educational values communicated to both males and females would be roughly the same, that the vocabulary and topics each is exposed to is roughly the same, that the level of family function or dysfunction each is exposed to is the same, that the economic conditions of each is the same, etc.
The question becomes, what could possibly be so significant as to account for those significant differences? It seems doubtful that those features will turn out to be so easily identifiable and quantifiable that they could ever be subjected to a truly scientific analysis.
And this indeed may be the fate of environmental influences more broadly speaking -- as is indeed suggested by the larger mystery of the Flynn effect.
It is not taboo to consider intelligence variations. It is taboo to take outcomes driven by factors other than race and make specious claims about racial disparities in intelligence. That is the major flaw of the Bell Curve and those who support it.
Kieran Healy has a nice succinct post on this topic as well. He summarizes it perfectly:
"In other words, the gap between blacks and whites on the only thing we can measure that might be related to underlying intelligence washes out when we take environmental factors into account! So where, exactly, is the controversy?"
In other words, the issue here is not whether the 'left' is unwilling to admit to intelligence differences between groups it is that there is no quanitfiable intelligence difference when other determinants are taken into account.
Comparisons to the measurable lack of 7 foot pygmy Jews or whatever is missing the point. As an aside, let me say that physical differences between races are also heavily influenced by environmental factors. Anyone who doubts that should visit Asia or Africa and compare the physical attributes of today's young with those of a generation or two older.
I repeat what I posted at Eschaton, modulo links (Aside to Bucky: don't miss that 80 year old stuff that not only can't Jews play basketball, they're stupid. Oh, but then in a generation they caught up and surpassed the white average on IQ tests?! Hard for you guys to explain, eh?)
1. Atrios is correct that CalPundit is much too nice to the authors of The Bell Curve. I suppose this was following the principal of interpretive charity, but he overlooks evidence (presented in the previous comments) that the book draws heavily from biased eugenicist sources, throughout.
2. Notwithstanding point 1., CalPundit has correctly identified (if I may simplify his argument) that a great many liberals rushed to drown out "The Bell Curve"'s racist conclusions not because they recognized why they were false, but because (for want of statistical and scientific training) they feared they were true. And this fear is why so much of the argument takes place at an ad hominem level, and not with any scientific analysis, notwithstanding the racists' failure to submit any of their work for peer review. One could usefully contrast this behavior with the liberal support of Evolution over Creationism, where, because the liberal side is comfortable with the quality of its scientific argument, it has erected a scientific superstructure that obviates the need to attack Creationists as the religious bigots they are. [Similarly the anti-Nazi site here]
3. I do regret that CalPundit didn't include S. J. Gould's one-sentence refutation (from "The Mismeasure of Man") of the sorts of racist blacks-are-hereditarily-less-intelligent argument. Seventy years ago, Gould explained, the differences in height within Japan were mostly explained by hereditable factors, but the difference in height between Japanese and Americans was mostly explained by environmental factors (principally, diet). And we can see the truth of this statement, as Japanese grow much taller as their diet and our diet converge. Genetic factors, Gould shows, are not reliable in examining differences across groups.
4. So, liberals, go read Gould and relax. When we just yell, "Racist, racist," the bad boys can smell our fear. When we yell, "Untrained, incorrect, incompetent flat-earth fraudsters", well, they don't change their racist minds, but they're not winning the battle for everyone else's hearts and minds, either.
>don't miss that 80 year old stuff that not only can't Jews play basketball, they're stupid. Oh, but then in a generation they caught up and surpassed the white average on IQ tests?!
Has the number of Jewish pro-basketball players, a proxy for the quantity of lithe 6'-6"+ Jews, risen in parallel?
If not, why not?
Memo to thread readers: References to Jewish NBA stars is intended more as humor than as deep analysis.
Gould and Lewontin also led a long, dishonest and losing battle against sociobiology and evolutionary psychology. The truth of the matter is that we have NO idea about the extent of the causal relation between genes and differences in cognitive abilities across populations. I'd say there is a good chance we may find out that genes do play a key role. Then what? Smart liberals should build their views of the world, ethical convictions and policy prescriptions such that IT SHOULD MAKE NO DIFFERENCE. One should not base one's ethical views on unproven factual statements (irrelevance of genetics on X, Y or Z) which can be falsified by genetic research.
BTW, I actually *read* The Bell Curve from cover to cover, and I disctinctly recall the authors claiming that heritability of intelligence couldn't be quanitified better than to say it was *roughly half* of the influence on testable parameters, the balance of influence being, of course, environment.
There are 500 people in the world play in the NBA. These 500 people are drawn from a small sample of athletes who
a) are physically gifted
b) play basketball instead of football, soccer, baseball, tennis, hockey, track etc.
c) decide to dedicate a large part of their life to athletics instead of academics.
How Bucky can make general inferences about the population at large from this sample boggles my mind. But then again, why am I surprised given the history of Bucky Dent on this site.
"Out of more than 720 [Nobel] prizes awarded since 1901, more than 130, or about 18 percent, have gone to Jewish laureates....Jews comprise only 0.3 to 0.5 percent of the world's population."
So, how do we explain overwhelming ethnic concentrations in obscure, demanding domains?
Identifying the right talent early, giving them better education, better mentoring, better guidance and better parenting and teaching. Also because of positive sorting of mates from the parental generation - smarter parents may produce smarter kids for a variety of reasons.
I don't think people opposed to the Bell Curve are arguing that everything is nurture not nature. What they are arguing is that the nature part is not conclusively distributed differently between ethnic groups.
To put it in your terms, are there any Ethopian Jews among the Nobel prize winners? No.
Is it because the lower intelligence from their dark skin color cancels out that their higher intelligence from having Jewish ethnicity? Or is it because the educational opportunities they and their parents had access to were far more limited than the Israeli jews?
>lower intelligence from their dark skin color
At some point, even you must limit the knowing, grotesque mischaracterizations of others' arguments.
Well, you made the following observation
* 18% of Nobel prizes have been won by an ethnic group that comprises only 0.1% of the population
From your prior posts, you seemed to be indicating that you believed this observed phenomena could be explained by a higher intelligence in that ethnic group.
I was pointing out that the distribution of Nobel prizes seems to be very different among memebers of that ethnic group with different racial or socio-economic characteristics.
We have two options to reconcile this intra-ethnic discrepancy:
a) It is not really race or ethnicity that matters but instead it is various characteristics related to the persons background or upbringing (this is what I believe)
b) Your claim about the superior intellect in the ethnic group is true in which case the only way the observed phenomena can be explained is by attributing a negative intellect to skin color. (the only way the observed phenomena is true given your belief)
So which is it Bucky? You can't claim that you believe in differences across ethnic groups in the positive dimension but then claim to have been misrepresented when the claim is analyzed in depth.
So if you are scandalized by the logical implication of your previously stated belief then it seems like you, like me, don't believe in intelligence differences between groups as a whole.
>you seemed to be indicating that you believed this observed phenomena could be explained by a higher intelligence in that ethnic group
Nope. Never said that. Never implied it. Clearly stated the BC's authors themselves attributed inherited components to be no better than *roughly half* of measured outcomes.
>Your claim about the superior intellect in the ethnic group
Show me where *I* *claimed* that. You can't.
Sowell, in his landmark book Ethnic America, convincingly ascribes the different life outcome distributions of various US ethnic groups to each cohort's culture/value system.
He tracks groups through generational time, chronicling how through the decades, such detested groups as Irish, Jews, Italians, Japanese and Chinese clawed their way into and through the middle class.
BTW, re: your reference to Ethiopia's Falasha Jews having no Nobel Prizes, I assume there aren't many native-born Ethiopeans in the NBA either. :)
John Maynard Smith, the "dean" of evolutionary biologists, says of Stephen Jay Gould that among evolutionary biologist Gould's ideas are regarded as too confused to be taken seriously. (Professor Smith said this in an article in the New York Review of Books.)
Gould made a big career for himself bowdlerizing Darwin to conform with political correctness. We all know the drill, we are all equal, there's no such as intelligence or it there is its due to environmental factors. Paul Krugman cites Smith's remarks and compares "pop economists" like John Kenneth Galbraith and Robert Reich (and the supply side boys at the other end) to Gould.
Every one is entitled to his own opinion on a scientific controversy. But the truth or falsity of a scientific theory should be treated as an empirical question, not a moral one.
That last sentence by the way is the opinion of Professor Noam Chomksy. Professor Chomsky doesn't think Herrnstein and Murray prove their case, but his line is that it wouldn't change his political beliefs if the Bell Curve were proven to be right, and that the question is empirical rather than ideological.
By the way, R.C. Lewontin is a Marxist-Leninist who does not believe in orthodox western biology.
Okay the basic point is that you disagree with the Bell Curve guys then. That's good to know.
By the way, all your oblique posts about not reflexively condemning the BC, the lack of 7 foot Jews in the NBA and the overwhelming presence of Jewish Nobel Prizes were not exactly making that basic point very clearly.
As a paid up member of the Flogging a Dead Horse Society (we think it important to keep flogging because these corpses have a nasty habit of resurrection), I think we should build our whips from a few simple points:
1) Yes, of course its OK to hypothesize innate differences in intelligence between groups. But hypotheses need honest, proper testing;
2) Given that different human races evolved very, very recently and AFTER the explosion in brain size, our priors must be that the hypothesis is unlikely to be true;
3) Given the known huge variation in IQ WITHIN groups, that environment has SOME impact on intelligence, and that environment varies massively BETWEEN groups, it will always be a very difficult hypothesis to test;
4) It is unlikely, therefore, that the posterior probabilities after simple tests ought to be much different from our priors.
5) Finally, the policy implication of (3) is that innate inter-group differences in IQ have no policy implications - their group membership tells us little about the INNATE IQ of an individual, even if this does differs between groups (because it's swamped by the variation within the group and by the between-group variation in environment). It is therefore more efficient (yes, as well as equitable) to assess each individual's innate ability on other evidence than their group membership.
Andrew Lazarus' point is true - simple outrage against the Charles Murrays of this world is bad strategy . You need to demonstrate their incompetence or, in Professor Murray's case, their bad faith (I'm picking on Murray here because I KNOW he's dishonest in a field I have professional expertise in - the economic effects of the welfare state). Its not very hard.
I had to comment on the Bell Curve post. Two years ago, Flynn and Dickens published a paper in Psychological Review on the IQ paradox. The IQ paradox addresses the question of why IQ increases from generation to generation. Their answer is that genetics is important and environment is important, but that there is a very strong environment X genetic INTERACTION. Thus studies that show that 75% of intelligence is related to genetics are lumping the genetic component and the environment X genetic components.
A layman review of the article is on the Brookings site:
Those that can make it through the mathematics should consult the original article.
Joe Willingham, would you mind posting your biology credentials so we can evaluate your second-hand ascriptions of JM Smith and SJ Gould?
I agree with Deridda Derider- Charles Murray and other beneficiaries of right wing foundations like Bradley cannot be effectively challenged by outrage alone. Dickens and Flynn have developed a model of learning and intelligence that is consistent with the IQ data and flatly refutes The Bell Curve. The Bell Curve depends on data that intelligence is primarily due to inheritance. However, this fails to account for the rise in IQ scores over time. It also overlooks the fact that considerable interbreeding between races has taken place in the US and that the white population contains numerous individuals with black heritage and the so-called "black" population contains significant numbers of individuals with European ancestors.
The reliance on the correlation between inheritance and IQ has been addressed by Dickens and Flynn (see post 23). Dickens and Flynn present a model that examines the INTERACTION between inheritance and environment. If one looks at correlation alone, then inheritance can account for 75% of IQ. However, such analyses assume apriori, no interaction between inheritance and environment. These analyses then are NOT measuring the contributions of inheritance ALONE to IQ, they are measuring the contributions of inheritance PLUS inheritance X environment INTERACTIONS.
Dickens and Flynn use a basketball analogy to present their case. A person who is just a little bit taller, or quicker than average is predisposed to be a better basketball player. A slight advantage may lead that person to play basketball more often and assoicate with others who play basketball. In the Dickens and Flynn model, a small genetic advantage can lead to a set of interactions that have a "social multiplier" effect. Being genetically predisposed (height, speed) to basketball leads to social interactions (environment) that improves basketball skills. In the sports arena, athletes that strive to be elite do so by playing with other elite players and learning from the competition. Dickens and Flynn use the basketball analogy of being able to shoot with both hands. 50 years ago, almost no basketball players did this. However, as players learned to shoot and pass with both hands, it gave them an advantage and put them in the elite status. In order to make the top grade, everyone who aspires to that level must learn to use both hands. Thus over time, basketball skills improve so that today's players are better than past players.
In education, the same type of model also applies. Colleges bring students together with faculty to interact because it creates a LEARNING ENVIRONMENT. Students learn by listening to lectures and reading texts, but that learning is greatly reinforced by study groups and interaction with other students that are interested in the same subjects. Thus students that are genetically predisposed to doing well in class, are good readers, etc. can gain admittance to a social structure that supports and reinforces learning through the social multiplier effect. Dickens and Flynn therefore argue that relatively minor differences in inheritance can lead to very large differences in IQ through the SOCIAL MULTIPLIER effect.
The Dickens and Flynn model and the Social Multiplier effect have a great deal of support from academia. Their model does a good job of explaining why temporary enrichment programs often fail. They work well when the social multiplier is in effect. Once the social multiplier is removed, then the participants sink to their innate levels. The Bell Curve is clearly on the wrong track. Environment is key to intellectual development. It is important to focus on environment and how to create environments that develop the potential of our youth. The Bell Curve would have us give up on the potential of our youth as the fate of inhertitance. Dickens and Flynn suggest that exposure to appropriate social multipliers can have a positive effect on students. Do we give up as The Bell Curve Suggests? Or do we put our efforts into finding strategies that bring the social multiplier to bear on the education of our youth?
It's been a long time since I've seen a discussion of "The Bell Curve"--I thought it had long since taken its place with "The Turner Diaries" and "The Protocols of the Elders of Zion" as a staple at gun showns and a souce book for Rush Limbaugh, but pretty much discredited elsewhere.
Does anyone remember the rather blatent statistical fraud that the authors attempted? Although every other statistician in the civilized world always states the regression coefficient in terms of r**2, Murray and Hernstein use r instead. This has the result of making there correlations look much, much stronger than they really are (0.5 instead if the true 0.25, etc.) One you recognize a few of their tricks, it becomes obvious that the entire "scientific" part of the book is a crude scam.
IIRC Maynard Smith's assessment of Gould was not directed at Gould's 'political correctness'. Rather, he was explaining why so many evoutionary biologists who disagree viscerally with many of Gould's best-loved hobbyhorses - punctuated equilibrium, the strong role of contingency, rejection of pan-adaptationism, a suspicious readiness to entertain group-selectionist notions and other shots cocked at the 'hardening of the [20th c. neo-Darwinian] synthesis' - tended nonetheless to avoid slagging him, at least in non-specialist fora.
His answer was that Gould was doing yeoman work against the creationists; attacking him would merely supply them with lots of juicy quotations to be used, misleadingly and out of context, in their efforts to spread darkness. From this view it was Gould's professional opponents who were being 'politically correct'. That's unfortunate, though one certainly has sympathy for their motives.
Before "Herrnstein" meant a thing to me, I had a chance conversation with him. Walked out of his office thinking this is the screwiest guy I ever met. Happily, I never met "Murray" because watching him on PBS was watching self-parody. Sure, ad-hom bad-hom. Nuts is nuts.
I went back to the BC summary I linked, and found this nugget: "The one intervention that has consistently worked to raise intelligence is adoption form a bad family environment into a good one."
I think that calls into question the assertions that the BC is a racist/eugenics tract that dismisses the learning environment as an input.
Just to clear up this mystery once and for all. Could you clearly articulate which of the following statements you think the Bell Curve demonstrates. Also it would be nice if you could articulate which of these statements you believe in because frankly your postings are confusing.
a) There is a strong link between environment/culture and intelligence
b) There is a strong link between race/ethnicity and intelligence, independent of environment/culture
c) Both a) and b)
The phrasing makes the answer a snap. A.
Cool. Glad to know you, like many of the rest of us, don't believe in b) or c). I think I misunderestimated you Bucky.
Question: Why is the Bell Curve suddenly re-appearing as a topic of discussion in blogland?
Mrs. Tilton writes:
"IIRC Maynard Smith's assessment of Gould was not directed at Gould's 'political correctness'. Rather, he was explaining why so many evoutionary biologists who disagree viscerally with many of Gould's best-loved hobbyhorses - punctuated equilibrium, the strong role of contingency, rejection of pan-adaptationism, a suspicious readiness to entertain group-selectionist notions and other shots cocked at the 'hardening of the [20th c. neo-Darwinian] synthesis' - tended nonetheless to avoid slagging him, at least in non-specialist fora."
That is correct in regard to John Maynard Smith's New York Review of Books article. But in his book *Shaping Life: Genes, Embryos, and Evolution*, Prof. Smith discusses the political dimensions of orthodox Darwinian biology versus the holistic school, of which Gould was a member. The holists reject the traditional "reductionism" of western science. Prof. Smith relates his own movement away from Marxism to his conversion to orthodoxy in biology.
Many writers noted the political slant of Gould's work at the time of his death. Those on the left praised Gould for the "progressive" bent of his work, while those in the center and on the right deplored the way Gould tried to rewrite biology to conform to the canons of political correctness.
If I'm not mistaken, to the degree they used r^2, they interpreted it as one would in a linear regression rather than in the logistic models they were actually using. Altogether pretty sloppy.
IMO, however, the real failing of the models were the crude scoring measure of SES and the treatment of success outcomes as binary (e.g. being above, as opposed to below, the poverty line constitutes a success, a rather simplistic picture). Of course people do this sort of thing all the time, but not when using their work to bolster such an outrageous social agenda.
Oh, and the "middle class values" index was also pretty strange.
"Question: Why is the Bell Curve suddenly re-appearing as a topic of discussion in blogland?"
Second-order effect from the Michigan affirmative action case, I'll bet.
Well, as the only person I know who's actually read the book, I can vouch for the fact that the chapter on race is not the entire book, although of course that's all the critics who haven't read it fixate on. ANd the authors do give a fair amount of time to environment.
Here's the problem: you can substitute "intelligence of the parents" for "environment". Whether those parents acquired their IQ through heredity or environment is irrelevant; they pass it on to their children. Low IQ parents don't read, watch more TV, don't play as many games or answer as many questions, etc. I'll leave it to others to argue chicken and egg. Efforts to make that up with early intervention programs don't seem to work: head start is extremely intensive, seems to have a rigorous methodological foundation, and shows only very moderate short term effects which disappear by the time they're in 3rd grade. There is no way anyone's found to substitute a teacher for the intensive, time consuming, one-on-one participation of a parent. So saying "it's the environment" doesn't resolve the unsettling questions raised by the persistent gap; it just raises new ones.
Incidentally, a friend who does this kind of research tells me that that study from the 20's showing that Jews scored lower than the rest of the population was a myth; it either didn't exist, or had some kind of methodological flaw, I can't remember which. Jews in the 20's tested significantly higher than the rest of the population, just as they do now. Whether that's because they live in a culture that is intensely learning-centered, or because their genetic heritage is better, I won't venture to speculate.
"Here's the problem: you can substitute "intelligence of the parents" for "environment"
I think the studies about adopted children show that substituting "intelligence of the parents" for "environment" is not straightforward at all.
Anyway, no point wasting my breath discussing the merits of a book whose methods have been pretty thoroughly debunked by Jim Heckman, a man far better equipped to criticize such work.
Michigan case aside, the book will continue in its rightful place in the trash-heap of history
For a while there it looked as though the theory of multiple intelligences had fought a battle against "intelligence is a unitary quality which Gerorge W. Bush doesn't have", and lost. Is MI making a comeback, or is it only conservatives like Sowell who believe in it now?
No, the twin studies seem to show that approximately 50% of the variation between identical twins is heredity, with the rest environmental. Of the 50% non-hereditary, the majority of the variation is attributable to intelligence of the parents, or close proxies like education. (The effect of socioeconomic status is apparently surprisingly weak after you take out educational attainment and IQ of the parents.) Or so I'm told by people who study these things. The environmental factors that we as a nation are willing to control, such as schooling, make up a very small fraction of the overall varation in intelligence. Nor is it obvious how flattening socioeconomic variation would help, even if the problem were the poverty itself rather than correlated factors such as education, since the problem with the American poor rests more on lack of status than lack of books or malnutrition. We'll still be primates no matter how many laws you pass against it.
It seems unlikely that you could produce a major change in outcomes with the sorts of interventions a liberal democracy is willing to stage, regardless of the race of the people involved. The town my mother is from has housed for the last 100 years the state school for the mentally retarded. When the high-functioning retarded leave, get married, and have children, their children have an extremely high likelihood of also being retarded. It's a small town; there's no significant environmental variation beyond the family. I have no idea whether or not this is because of genetics or environment, but if it were the latter, are we prepared to remove the kids from the home and put them into adoptive families that can read to them, take them to museums, play games that provide a high degree of mental stimulation, etc? I think having the state judge who is intelligent enough to be a parent is repulsive. But if not, how do we rectify the lack of stimulation at home? Even if we were willing to commit the resources, where are you going to find the number of adults necessary to provide the hours of mental stimulation that high-functioning parents give their children? And what should we do about parents who choose not to enroll their kids in the extra programs, as parents do now with Head Start?
I think you are confusing the argument "there are no measurable differences in intelligence across race/ethnicity once you control for other factors" with "there is no hereditary component in intelligence once you control for other factors".
I believe in the first, I don't believe in the second; although I think much of the hereditary effect can be washed out by subsequent behavior. As I stated earlier, I don't believe that nurture can explain everything, nature has a role to play as well, although the strong correlation between the two makes untangling effects very different.
What is offensive about the arugument of Bell Curve proponents is not the fact that they claim Nature has a role to play. It is that they claim that Nature has a role to play in a specific way across different races and ethnic groups. It is a claim that as Heckman and Sowell show, that the BC authors fail to prove spectacularly, but once that is very dangerous given how it can and has been used in the past.
As I said it in a prior post before, Kieran Healy's summary still stands as the best summary of the 'controversy'
"In other words, the gap between blacks and whites on the only thing we can measure that might be related to underlying intelligence washes out when we take environmental factors into account! So where, exactly, is the controversy?"
"So saying "it's the environment" doesn't resolve the unsettling questions raised by the persistent gap; it just raises new ones."
What are these "unsettling questions?"
Even though I don't know what these "unsettling questions" are, it seems to me that most "unsettling questions" would be rendered irrelevant, if:
1) IQs increase over time, such that that the IQ of the lower-IQ group is equal to the IQ of the higher group of only 1 generation earlier, and
2) Humans aren't going to be the smartest group on earth merely 20 years from now, anyway...machines will be. (In fact, machines are already smarter than humans in many tasks, such as stock trading. I'm speaking of the time when machines will be smarter than humans at most tasks.)
"And what should we do about parents who choose not to enroll their kids in the extra programs, as parents do now with Head Start?"
Shoot them? ;-)
Put them in re-education camps? ;-)
Sorry, I tend to think first of ideas that have been tried in the past. ;-)
"It seems unlikely that you could produce a major change in outcomes with the sorts of interventions a liberal democracy is willing to stage, regardless of the race of the people involved. The town my mother is from has housed for the last 100 years the state school for the mentally retarded. When the high-functioning retarded leave, get married, and have children, their children have an extremely high likelihood of also being retarded."
Is this the sort of "unsettling question" you're thinking of?
It doesn't seem to be particularly relevant to the questions raised by potential/actual difference in IQ among "races."
First off, the difference in IQ among "races" seems to be so small that we can't all even agree whether or not they exist. It's hardly on the same level as the difference between a person of median intelligence and the "higher functioning retarded" persons of your example.
Second, even if these small differences exist, they hardly seem to have any effect on the overall performance of world groups: 1) we kicked the Empire of Japan's butts in WWII, 2) the mainland Chinese were a basket case, until they started to abandon socialism, 3) the Jews hardly rule the world (the paranoid fantasies of some notwithstanding), and 4) the barriers to interacial marriage are breaking down all around the world, and are likely to continue to do so (unless there's some sort of world-wide insanity, such that people like Tiger Woods and Halle Berry are no longer considered attractive potential mates).
So, again, I don't really see the serious potential problems raised by a potential difference in average IQs among "races."
There is no comfort in the IQ data for white racists. European Jews as well as Chinese and Japanese score considerably higher than non-Jewish whites.
This is the case whether Murray and Herrnstein or their critics are right.
>machines are already smarter than humans in many tasks, such as stock trading.
This is simply and totally false. I work with computerized trading systems for a living. The statement is nonsense.
Computers have to be elaborately programmed by humans to do their work, and even then they aren't "smarter", only faster.
"Computers have to be elaborately programmed by humans to do their work, and even then they aren't "smarter", only faster."
I agree with Bucky. Humans have to design the models, the computer merely carries out what the model tells them to do.
That's just the point. Race was one or two chapters of a very large book; the debate is much wider than whether or not there are heritable gaps across the races.
"This is simply and totally false. I work with computerized trading systems for a living. The statement is nonsense. Computers have to be elaborately programmed by humans to do their work, and even then they aren't 'smarter', only faster."
Oooh! I hit a nerve! ;-) My apologies for any offense given. But my statement was hardly "nonsense." I think it's essentially what Ray Kurzweil has said on several occasions...including in "The Age of Spiritual Machines."
In the field of artificial intelligence, Dr. Kurzweil doesn't say "nonsense." He says things that are "controversial," perhaps. "Wrong," even. But I can assure you that he's given a great deal of thought to the matter of artificial intelligence!
The central question is how we define "intelligence." For example, several decades ago, many artificial intelligence experts thought that computers would never be able to beat a Grand Master at chess, because computers would never be "smart" enough. Now, when computers *can* beat the best human chess player in the world, the human consensus is now that chess playing isn't "smart"...at least in the way that the machine does it. A dumb machine can be programmed to beat a human.
But the question is, what is "intelligence"? If you define it as "a machine that beats humans at X" (where X can be chess playing, stock trading, plane flying, etc.), then, by definition, the computer program that beats a human is "smarter" than the human.
But if you define "intelligence" in some other way (i.e., "can design computer programs that beat humans in chess playing, stock trading, plane flying, etc.") then computer programs aren't there, yet.
Kurzweil prefers the first definition, because it's simpler. I agree that the first definition is simpler, and therefore probably better; words are better when they mean one thing, consistently.
You obviously choose something like the latter definition for "intelligence" (understandably). Using the second definition, computers aren't even close to being as "smart" as humans in stock trading...though they probably will be, by that second definition, within the next couple of decades. Ray Kurzweil will probably help see to that.
"That's just the point. Race was one or two chapters of a very large book; the debate is much wider than whether or not there are heritable gaps across the races."
Was this directed at me? If so, I still don't understand what the "unsettling questions" are. I understand how the question of retarded people having children could at least *seem* unsettling (though, as long as they don't have dozens, which seems unlikely, I don't see a real significant problem).
What are some (other) "unsettling questions"?
Since this thread has now cleared 50 posts, i wish to point out that none of the last 48 posts changes the pithy conclusion that jd reached in post number 3. In his immortal words:
The Bell Curve is rubbish."
>machines are already smarter than humans in many tasks, such as stock trading....Oooh! I hit a nerve!
No, you didn't hit anything. You simply made a statement that I know is flat-out untrue. Show me a "machine" that's outperformed the market over a cycle witout human intervention/reprogrammng, and I'll admit I'm wrong.
"Since this thread has now cleared 50 posts, i wish to point out that none of the last 48 posts changes the pithy conclusion that jd reached in post number 3. In his immortal words:
"Aaaaaaaaaaaaaaaaaugh. The Bell Curve is rubbish.""
Well, that conclusion is worth the paper it's printed on. Absent supporting arguments (WHY it's "rubbish"), an opinion isn't worth much.
"Show me a "machine" that's outperformed the market over a cycle without human intervention/reprogramming, and I'll admit I'm wrong."
Well, any comparison of *current* machines against *current* market performance is comparing machines against other machines.
If you look at the second article I linked to, you'll see:
"Hamilton's investment instincts beat the market by nearly 3 percentage points a year between 1930 and 1997. There's just one hitch–Hamilton died in 1929. His results are real, but he is not–at least not any longer. Those sparkling returns were produced by a VirtualHamilton neural network–a branch of artificial intelligence whereby software programs "learn" through trial and experience–created by a team from New York University and Yale. The real Hamilton ran the Journal in the early 1900s, but the academics mined his writings to replicate the pundit's mind. They then fed the VirtualHamilton seven decades of market data to see how it performed."
Those are data comparing a machine with the market, over a period when machines weren't used (at least up to the very end).
That machine beat the market by 3 percent per year, over a period of 67 years. Sounds pretty good to me!
>VirtualHamilton neural network–a branch of artificial intelligence whereby software programs "learn" through trial and experience–created by a team from New York University and Yale. The real Hamilton ran the Journal in the early 1900s, but the academics mined his writings to replicate the pundit's mind. They then fed the VirtualHamilton seven decades of market data to see how it performed."
This describes a training function where a network "learns" to profit by adjusting its rules to maximize profit over an **existing static** data set. Almost all such efforts show a "profit".
But, that's only the step you take before using the network thus created **forward** on dynamic reality.
To be fair to you, *I* would not have known the difference had I not been involved in this field professionally for several years.
"This describes a training function where a network "learns" to profit by adjusting its rules to maximize profit over an **existing static** data set. Almost all such efforts show a "profit"."
"But, that's only the step you take before using the network thus created **forward** on dynamic reality."
This doesn't sound any different from what I thought it was. They set up the model, based on Hamilton's writings (including, obviously, Hamilton's commentaries on the market, and market performance) prior to his death in 1929. Then, they compared the results of that model against the market, after Hamilton died.
I still say that their model (machine) outperformed the market.
I have an analogy, based on work I've done. I developed a model to predict styrene emissions from fiber-reinforced plastics fabrication (like tub and shower stalls, spas, truck caps, fiberglass boats). The model was based on the results of an initial set of research tests.
People would then say, "I run a tub and shower facility, and use such and such materials...what do you think my emissions are?" I would tell them, "Don't go by what I think, go by what the model says."
I could have probably made a pretty reasonable guess at what their emissions were, since the model didn't have too many variables, and I'd set up and run the model. But the model was "smarter" than my guesstimate of what their emissions would be.
Of course, the users could also put in some really bogus inputs to the model, and produce results that I could definitely beat with my guesstimates. So one could also say that the model was pretty stupid.
Also, when we did some later testing, to refine/correct some aspects of the model, the model itself couldn't incorporate those results. I had to change the algorithms myself, based on that new testing. So the model was stupid in that way, also.
But overall, in my opinion, the model was "smarter" than me, in predicting styrene emission factors from fiber reinforced plastics fabrication. Given various parameter inputs, I would only have a guesstimate of what the emission factor would be.
From Ray Kurzweil's site:
"Machines with sharply focused intelligence nonetheless became increasingly pervasive. By the early 1990s, we saw the infiltration of our financial institutions by systems using such computerized techniques as neural networks, markov modeling, time-delay embedding and fractal modeling. Not only were the stock, bond, currency, commodity, and other markets managed and maintained by computerized networks, but the majority of buy-and-sell decisions were initiated by software programs that used these techniques on increasingly extensive on-line databases (e.g. Midland Global Markets' 16 Gigabyte database in 1993 recorded every moment of trading in the Standard and Poor 500 and several other markets for the prior 10 years)."
I agree with what I recall to be Dr. Kurzweil's position in The Age of Spiritual Machines, which is that, if a computer is used in preference to humans in some task, the computer should be considered "more intelligent" for that particular task. Since computers are doing the majority of buy-and-sell decisions, they are "more intelligent."
Both Dr. Kurzweil and I recognize that humans are setting up the neural nets and databases that allow these programs to work. Computers can't do that...yet. So one could also say that computers aren't yet as smart as humans in buying and selling stocks.
Markets are not physical systems with rigidly forecastable outcomes/equilibria.
Testing/training a network on an existing static data set is different than using the network to risk capital on an unknown present.
Computers can "make markets" in the sense that they can raise or lower bids/offers on related securities according to human created algorithms, but they cannot establish the intitial relationships between those securities, nor can they forecast changes in those relationships.
There are traders who use "black boxes" for specific functions -- like dedicated option buying or selling programs -- but, again, the computers merely efficiently execute instructions provided by their human masters.
Old saying: The models work, until they don't.
>Those sparkling returns were produced by a VirtualHamilton neural network–a branch of artificial intelligence whereby software programs "learn" through trial and experience...
This is the key phrase. Hamiliton's notions were likely the intitial model weights, which then were turned loose on the market price data, where the network "learned" to optimize itself for maximum profit on the STATIC data set.
I've done this HUNDREDS of times. The models need to retune on dynamic real-time data in order to make money, and reality is sufficiently plastic and varying that the neural networks rarely keep pace for more than a few weeks at a stretch.
If you think you can beat the house, try these:
May the force be with you. :)
Well, there's a first time for everything, but I agree with Dent. Arbitage programs are at least a couple of order of magnitudes more difficult than chess.
“For example, several decades ago, many artificial intelligence experts thought that computers would never be able to beat a Grand Master at chess, because computers would never be "smart" enough. Now, when computers *can* beat the best human chess player in the world, the human consensus is now that chess playing isn't "smart"...at least in the way that the machine does it. A dumb machine can be programmed to beat a human.”
“Those are data comparing a machine with the market, over a period when machines weren't used (at least up to the very end).
That machine beat the market by 3 percent per year, over a period of 67 years. Sounds pretty good to me!”
A computer can indeed be programmed to deal with probabilities. And since the computer cannot become tired or nervous, it may even have at times a distinct advantage over a human being. The bloodless computer simply goes with the odds. It does not get sentimental or greedy.
The real challenge is to get the computer to think abstractly. Alas, no machine has written a poem or a novel. There will also never be a computer able to converse colloquially with human beings. Why am I so certain? That is due to the fact language is intrinsically nebulous.