Thanks for that post Collounsbury, and to others as well who have provided such excellent information and reasoned argument.
Save your rolleyes boy. In other words, you have a history of bullshit argumentation, of willfully and gratitiously misrepresenting data, citations, sources etc. as such your assertions to date on this subject get in the way of the discussion.
Passion is not the question. Accuracy is the question. Not willfully ignoring materials brought to the table, or misrepresenting them. In other
Why? I’ve carried enough goddamned water in genetics related debates, I simply want to see something approaching a non-distorted engagement.
Tomndebb -
I believe this book was written for the public. That’s why there’s a section of the book titled “Statistics for people who think they can’t learn statistics.” To say that it should have been subject to “peer review” first, like a purely scientific publications, is saying that public books of this sort should like scientific publications. If that is the rule with this book than why not the rule with all other books of like nature? It was this fact - that it was written for and made available to the public - that appears to have brought most of the mass of criticism it saw. Sadly, there appears to be those in our community who have assigned themselves as guardians of what information is or is not appropriate for public consumption.
More to the point – what errors are in the book are you talking about that make the general conclusions of the book invalid?
Tigers:
The problem as I understand it is this:
If you are writing a book for the public based on generally accepted science and theory, then that’s one thing.
If you are presenting a new scientific conclusion, that is another thing.
In the latter case, good scientific method is to offer your data and conclusions up to other experts so that they can check and challenge it.
To be honest, I am not quite sure what merited a response. I didn’t know what Spearman rho was a few minutes ago, so I looked it up. Everything seems to indicate that the equation gives legitimate product-moment results, though with my nonpareil background, I might not know the difference one way or another.
The more relevant question is what M&H are including as data sets X and Y. Apparently it’s all in the appendices. Which I do not have.
Collounsbury -
Since you apparently are unfamilar with the monozygotic twin studies – and as you stated “too tired” to look them up I’ll get that information for you. As to the “uninformed blithering” - it seems you just about admitted such.
[bTigers2B1**: "I believe this book was written for the public. That’s why there’s a section of the book titled “Statistics for people who think they can’t learn statistics.” To say that it should have been subject to “peer review” first, like a purely scientific publications, is saying that public books of this sort should like scientific publications."
Tigers, this is an incoherent argument, the false premise of which is that a book can’t be, on the one hand, peer reviewed, and on the other, written for the public. There is nothing to prevent a book from being both peer-reviewed, and written with a general audience also in mind.
“Sadly, there appears to be those in our community who have assigned themselves as guardians of what information is or is not appropriate for public consumption.”
You do not seem at all to understand the nature of peer review, which is to protect the public from charlatanerie. We live in a very complicated world: a person may be an expert tax lawyer, a leader in the field, and not understand a thing about statistics, genetics, or IQ-testing. In what sense does that tax expert benefit from reading a book about the latter subjects that has not been peer-reviewed by those who know and understand the relevant research and methods?
If you were in trouble with the IRS would you go to a geneticist for advice? If there were a dispute about tax law would you be content to rely on a book that had not been peer-reviewed by other lawyers?
“what errors are in the book are you talking about that make the general conclusions of the book invalid?”
Tom is more than equipped to answer for himself, but it seems to me that he has offered several examples of what “makes the general conclusions of the book invalid.”
I’m not presenting it as an ‘either / or’ argument. That is, I’m not saying that it can’t be both. I’m just saying that your argument would prevent all science based publications from being offered to the public without someone elses review first. Surely your not saying that. Now if all of this is NEW science as Scylla suggests that is different and can certainly understand that position.
Do you engage in anything but ad hominem attacks? I know you have a little fan-club going, but how about you try to make it through one OP without insulting someone?
This is true. On page 118 it says, “In the chapters that follow [i.e., Part II, Cognitive Classes and Social Behavior], it [the NLSY] will play the central role in the analysis, with other studies called in as available and appropriate.”
Right. In fact, they refer to this study as the “mother lode for scholars who wish to understand the relationship of cognitive ability to social and economic outcomes.”
As was mentioned by someone, this study uses the Armed Forces Qualification Test as its measure of cognitive ability. They describe the AFQT as “a good measure of general cognitive ability,” citing their own Appendix 3. Appendix 3 goes into considerable detail arguing the validity of the AFQT as a test of g. Incidentally, I notice that Appendix 2 discusses NLSY in some detail.
Is TBC correct about the value of NLSY and AFQT? I dunno. They mention prestigious organizations, like NORC, and they seem to use the right words, but I am not in a position to confirm or deny this point.
Gadarene had asked, “Suppose the findings of The Bell Curve were and are 100 percent accurate–what would you consider to be the prescriptive implications? What should it mean for public policy, etc.?” TBC discusses policy issues in Part IV. Broadly summarizing their suggestions (after a very quick glance) they offer [ul][]Failures are apt to have low IQs, so programs for welfare recipients, teen-age mothers, criminals should deal with the likelihood that these people will have lower-than-average IQ’s. []Efforts to increase cognitive ability by altering the social and physical environment have had a spotty record, although they may succeed eventually.[]Education programs, such as Head Start or Vouchers or Affirmative Action, should be judged on their effectiveness. []Amercan education has been more successful with average and below average students, but has neglected the gifted minority.[/ul] These conclusions don’t seem particularly controversial…
That’s the best post I’ve read in a while, unabashedly lame even by december’s standards. It’s hard to imagine you guys could continue to argue against the book’s merits considering that it’s ummm, long and stuff.
You really shouldn’t be giving this stuff up for free, december. Someone’s gotta write the ad copy:
It’s longer than Lolita, broader in scope than Origin of Species, and contains more footnotes than the combined works of Dickens and Flaubert (count em, you’ll see).
Here’s december’s quote in context. He was responding to a post that seemed to imply that finding an error automatically invalidates the entire work. I add this only because your selection of a portion of ** december’s ** statement misleads.
Here’s december’s quote in better context. He was responding to a post that seemed to imply that finding an error automatically invalidates the entire work. I add this only because your selection of a portion of ** december’s ** statement misleads.
** What, you’ve never heard of meta-analysis? 
Rather than attempt to swallow the nine hundred pages and copious footnotes of TBC whole, let me back up a step and lay out a simplified argument. For the moment, let’s try and work with this argument rather than TBC so we can, perhaps, identify where the sticking points are. I hasten to add that I’m merely an interested layman in this area and that it is quite a while since I read TBC. Some of the premises I’m laying out may seem a bit odd initially, but try and work with me here.
1. There exists a set of standard tests. A person’s score on these tests is correlated to some extent with his or her general level of success in life.
Note that I don’t say that there is a thing called “g” or “IQ” or whatever.
2. There is a genetic component to a person’s ability to score well on these tests. All else being equal, people who’s parents scored highly on these tests are more likely to score highly on these tests than people who’s parents did not score highly.
Just to be clear, I’m not saying that environment isn’t important. It clearly is, especially in a negative sense. While it’s probably unlikely that a “perfect” environment can turn anyone into a genius, it is undeniably true that sufficiently bad environmental factors can turn anyone into a literal moron.
3. There is a significant difference in average scores on these tests between different arbitrarily defined population groups.
I think this is a key point. I think it is somewhat of a red herring as to whether or not “white” or “jewish” or “black” or “east asian,” etc. are really “valid” population groups. My reasoning is that you’re using the same arbitrarily defined population groups to analyze IQ and social success/pathology. Now correlation is not, of course, causation. However, it is still statistically meaningful to say "There exists a group with some characteristic X. The average value of characteristic Y is higher for this group than it is for the general population. Characteristic Y is highly correlated with income. Therefore, all else being equal, the group with characteristic X will have a higher average income than the general population.
Therefore, it is likely that at least some portion of the difference in average social success between population groups has a genetic basis.
This is a likely sticking point as I can already hear people saying that this hasn’t been proven. Yet, given that the ability to score highly on these tests does have a genetic component, which is the more likely initial hypothesis – That exactly 100% of the difference in performance between groups is due to environmental factors or that some portion of the difference between groups has a genetic basis?
Note that this does not presuppose why these different groups may have a genetic difference. It does not follow that East Asians in the U.S. are “smart” because they are East Asians. Perhaps it’s a founder’s effect. “Smarter” East Asians were more likely to emigrate. In any event, if there really is a correlation, the reason for the correlation doesn’t really matter for our purposes.
At the end of the day, what does all this mean? Probably not a lot. One standard deviation on an IQ test is 15 points. An 8 point difference on an IQ test sounds like a lot, but it isn’t really, especially at the middle of the curve. If I did the math right, this means that if population A scores half a standard deviation lower than population B, only about 4% more of population A falls into the “sub-normal” range (-2 standard deviations from the mean) than does population B. In practical terms, this difference, if it exists, tells you nothing whatsoever about anyone you see on the street.
So where are the controversial points in this argument? None of them seem particularly outre’ to me, except the conclusion.
Colleunsbury
Here are some links to results of monozygotic twin studies which I said I would link. I doubt very seriously you will find any study from an .edu source that substantially differs from these. The monozygotic twin studies show a very strong correlation of genetics and IQ.
http://www.loni.ucla.edu/~thompson/MEDIA/NN/ns.html
http://www.sscnet.ucla.edu/anthro/bec/papers/Segal_Abstract.html
http://www.olemiss.edu/courses/psy529/Lectures/Beh.Genetics3/behgen3.htm
http://www.atlantis.edu/~nutmeg/neuro/twinstudies.htm
Colleunsbury
Here are some links to results of monozygotic twin studies which I said I would link. I doubt very seriously you will find any study from an .edu source that substantially differs from these. The monozygotic twin studies show a very strong correlation of genetics and IQ.
http://www.loni.ucla.edu/~thompson/MEDIA/NN/ns.html
http://www.sscnet.ucla.edu/anthro/bec/papers/Segal_Abstract.html
http://www.olemiss.edu/courses/psy529/Lectures/Beh.Genetics3/behgen3.htm
http://www.atlantis.edu/~nutmeg/neuro/twinstudies.htm
Another layman’s observation. Whenever someone sets out to prove “scientifically” that one group is “substandard” in an important human characteristic it probably isn’t because they have that group’s welfare as a high priority.
The Bell Curve appears to be a collection of carefully selected data, massaged in arcane ways in order to support a conclusion that was reached before the data were even assembled.
The book seems to be little more than a long winded exposition of the fact that its authors think group “A” is deficient as compared to group “B.” The only answer to the “so what” question, i.e. “What are the public policy implications of this?”, is to “simplify and clarify.”
Jeez, that took 900 pages?
Just a higher level of garbage in, garbage out.
Note that I don’t say that there is a thing called “g” or “IQ” or whatever. No, but you did say “general level of success in life.” This makes it hard to agree with this point. How about "explaining about 25% of the variance in school performance, 30% of the variance in years of education, about 25% of occupational status, about 15% of income, about 30% of job performance, and about 4% of criminality.
I largely agree, although can we say that there is an association between genetic factors and test scores? (Just a hair of a step down in the implication of causation).
Some groups differ, largely they do not.
It is likely that a portion so small as to be unworthy of discussion is related to social success. Consider the variance unaccounted for in the more clearly identified independent associations between IQ and later outcomes that I mentioned above.
Very good point.
It seems that the controversy would come from attempting to extend the argument, given all that is unexplained, to policy and societal decisions about the allocation of resources, hope and, what was GWB’s line about this, the racism of low expectations?
Scylla, I love the IQ Cap! Phrenology Returns! Too bad I couldn’t find anything anywhere else about it.
“2. There is a genetic component to a person’s ability to score well on these tests. All else being equal, people who’s parents scored highly on these tests are more likely to score highly on these tests than people who’s parents did not score highly.”
Actually, I think this a very weak prong in Truth Seeker’s argument. There is indeed a genetic component to a person’s ability to score well on tests, as with many other abilities. But it doesn’t follow that parents and children with a similar set of abilities–here test-taking–share these abilities on genetic grounds.
I am not exactly like either of my parents, and I am very different from one of my brothers. My son is neither a carbon copy of myself or my husband. I am much better at taking tests than either of my brothers; my son is much better at math than I am.
That said, all the adults in my family–mother, father, brothers, and husband–were good enough at taking tests to earn an advanced degree of some kind and pass some kind of professional qualification. As a result, all of us are socio-economically secure.
My husband and I read a lot of books. Surprise, surprise, my son reads a lot of books.
Do you think that’s because he has a gene for reading books?
Mebbe.
I like pop music; my son likes pop music.
My husband does a lot of work at home in the evenings and weekends. My son sits near him and does homework, reads books about science (a subject that my husband likes). My son watches little television at home. He goes to museums, planetariums, zoos, as well as the typical range of kid stuff.
His teachers tell me he has a good imagination; good concentration; good study habits; performs well on standardized tests.
My son is not adopted, but temperamentally he reminds me more of my grandfather than of either myself or my husband.
If my son were an adopted son, I doubt his temperament would remind me of my grandfather’s. He might be less good at math and better at something else.
But do you think it’s very likely that he would not like pop music, not work beside his dad, not have good concentration, not have good study habits and not perform well on standardized tests?
It is not parody; it is extrapolation. Murray and Herrnstein are fairly careful to make as few recommendations regarding education as they can. (Having dismissed education as a way to affect IQ, this is consistent, of course.
On the occasions where they do mention education, they are fairly clear that they don’t want it used to improve the lot of the disadvantaged. For example. they note that under Head Start, childrens’ IQ can go up as much as 10 points. Rather than recognizing that this mere introduction of education is damaging to their claim that it has no effect, they then point to the studies showing that Head Start participants undergo “convergence” in which their IQ scores are no longer different than their non-Head Start peers.
In other words, they acknowledge that intervention increases IQ, but then when the intervention is removed, rather than recognizing that IQ is subject to outside influence, they claim that the “convergence” is an argument against even bothering with intervention and that Head Start should be relegated to a glorified baysitter status that keeps kids in a bad environment saafe for a while, but is not used to try to provide better education.
A central thesis of the work is the use of the NLSY to establish the basic parameters of IQ (or g) and then to extrapolate those figures across all of society by correlating the NLSY numbers to the AFQT and using the AFQT numbers to demonstrate where the numbers “would be” if the NSLY had been administered to a larger section of the populace.
Using Murray and Herrnstein’s own figures, Christopher Winship of Harvard University and Sanders Korenman of Baruch College tried to replicate the results and found multiple errors. In other words, M&H were simply fudging the numbers in their core tool to extrapolate to the general population as noted in
http://slate.msn.com/default.aspx?id=2416&sidebar=50884
Claudia Krenz ( http://www.srv.net/~msdata/bell.html ) has also run an analysis of many of the statistical portrayals found in The Bell Curve. Interestingly, when she ran the tests on the AFQT, she also found that the numbers M&H produced were not reproducible, especially when applied to people living in poverty: http://www.srv.net/~msdata/analysis.html
This is not about making small errors among hundreds of examples as december would like to characterize it. Once they have completed their survey of the history of psychometry in the first half of the book (which is acknowledged even by detractors to be a lucid and even-handed exposition) in the second half of the book their central thesis requires that the NLSY be an accurate measure of IQ or g and that the figures they draw from that test can be expanded by the magic of the AFQT. Serious errors in their application of the AFQT pretty well demolishes the conclusions they draw from their extrapolations.