Race differences

[TCO]

Originally posted by Ramo


The first thing to note is that Murray and Hernstein have a clear political agenda for this book, which takes up several chapters by itself.

I agree.

1. They are upfront about stating their libertarian leanings.

2. They advise various policy recommendations.

Let's try to keep the "reports of IQ variability" seperate from the "what we should do about IQ variability". I know that somebody with your handle for multivariable calc can do this.

I'm going to respond to the various remarks that you made. But I think you should remember that when I referred to the Bell Curve it was as a source of study on IQ differences by race. (There is a lot more in that book.) I wonder if you have some specific issues with the study that they did using NLSY data. And if you can point me to studies that were more effectively done. If you think that was such a crappy study.

[TCO]

Originally posted by Ramo
(on a side note, he mentions that US immigrants, based on the tenuous data he has, should have lower IQs, without even mentioning Middle Eastern or South Asian numbers, at all ),

1. Agreed. The discussion regarding immigration is much sketchier than the extensive discussion of B/W differences. (I am looking at it right now. p.359-364.)

2. They are upfront about the limited nature of the knowledge on this issue. "Keeping in mind that we are hoping to do no more than establish a range of possibilities...."

3. Regarding ME and SE Asian numbers, it's not that H and M blow off considering them. They state explicitly that they don't have reliable surveys on these poulations. They do state that these groups along with "other" constitute 11% of immigrants in the 80's (Statistical abstract of the US, 1992 as source.) They state that they omitted these groups from the computation. Do you think they were slipshod and missed reliable surveys on these groups? Do you have some? I'm sure that they'd love to have them. If you put in your numbers from those surveys, how does it change the results? I'll leave it to you as a math exercise, to determine how much these groups would have to differ from 100 to change the direction of H and M's calculation.

[TCO]

Originally posted by Ramo


The authors almost completely ignore cultural factors. Their multiple regressions have very little relevance as they don't incorporate such biases (hence my earlier comment of IQ studies being inadaquete).

I'm really intrigued here. How would amend the regression? What specific factor would you isolate? Think about it. Or at least explain it more clearly to me. They did run SES as a seperate variable.

[TCO]

Originally posted by Ramo

The authors dismissed the scholarship out of hand (calling one researcher a "radical") that the existence of a "general intelligence" is rather tenous.

1. This touches on a different issue here. Is there such a thing as general intelligence? Does IQ measure it? So what do you think?

2. The radical school is self-identified as such.

3. The radical school refuses to try to quantify and test the "multiple intelligences" idea. If you can't measure it, it's not science imho...

[Tingkai]

And I thought I was bad with my multiple posts.

[TCO]

Originally posted by Ramo

They use absurd sources for a lot of their data. First, there's this Lynn fellow, who the authors rely on for much of their data. He happens one of the editors of the "Mankind Journal," a nice little ultra-racist publication that the Southern conservatives dumped their money into to try to preserve segregation, and a staunch supporters of Apartheid.

1. If your comments are completly true and fair, that would make me look very skeptically at data from him.

2. I beleive that he wrote a review* article or book. Rather than performing studies himself. But I haven't actually read any of his stuff. Have you?

As bad as his credentials sound, his data is even worse.

I particularly was amused how he used IQ tests mostly from South Africa under Apartheid (there was even one sample from the Belgian Congo) to represent the entire continent!

1. Can you suggest another author who has done a better review of IQ testing in Africa? Since you think that Lynn is biased and slipshod.

2. Are there some specific studies that he omitted? That support your point of view?

Lynn also looks at a cognitive test, "Progressive Matrices," (supposedly less subject to cultural biases) the creator of which has said that there is no meaningful translatation of these scores into IQ scores, in that the distribution is not normal. But Lynn normalized the data, despite knowing that a standard deviation means absolutely nothing.

This is pretty interesting. I've actually done some work looking at these kinds of things in job performance talent surveys. What was the shape of the distribution curve in the Progressive Matrices test? Degree of skew?

Of course, Africa isn't the only continent that he butchers. There was one source where the he took a sample of 58 Japanese kids, and compared those scores to a sample of several thousand American kids taken over a decade earlier. In another East Asian sample of IQ scores, scores were supposedly taken from "typical" students, but were in fact taken from economically well-off students.

1. I'm not sure since I haven't read Lynn** (...and I bet you haven't either! ), but one possible reason for some of these complaints is that Lynn ran a meta-analysis...so that he actually looked at several different surveys...many of which may have flaws. This is a Bayesian method.

2. Did these flawed studies make it into The Bell Curve? Or is this just more stuff to show how stupid Lynn is?

*Do you know what I mean when I say a review article in technical literature? It is a compendium of various experimental papers. Usually has a long bibliography. Sometimes has meta-analysis. Typically looks at trends among the different studies.

**But if you want, we can both look at his work. Should be easy for you to access since you are at a premier research uni. We should have some specific objectives/tests in mind when going over his work. probably would also be good to see if there is other work that surpasses his.

[TCO]

Originally posted by Ramo


Another tenous facet of Hernstein and Murray's book is the part on "reaction times."

I agree with you that this portion of the book is tenuous. Are we just arguing about Bell Curve in general, now? This specifically is not IQ testing, which was what I recommended it to you for.

[QUOTE] They were saying that asserting that whites have a faster "reaction time," that is the time it takes to make a decision (i.e. a signal getting to the brain), while blacks have a faster "movement time," or the time it takes to execute the decision. The study they

cited which supposedly gave "consistent" results in fact showed that both assertions were true for 2 out of 3 tests. What's more, Jenson, the person who created the study had tried it twice before. Interestingly enough the three studies were inconsistent.

That's very interesting. Did you look at the Jensen studies? What do you mean by 2 out of 3? like 3 people or 67% of a test group? Just not sure what you're driving at. Could you also explain what the situation was with the different iterations of the study? Not sure I get your point. Were there different papers (one for each study?) or different substudies in one paper?

I could go on and on about this kind of thing (check out "The Bell Curve Debate" if you're still interested), but the crux of the matter is that these studies aren't created by scientists, but racists hinding behind a mask of science.

I've read one compendium of criticism of The Bell Curve as well as reading The Bell Curve itself. In general, TBC is more scholaraly, has more data, more references to literature, is more even-handed in treating possible alternate explanations than the criticisms are. (I base this on having read about 40 papers in journals like J. Appl. Psych. and J. Labor Economics for an internal company research project and on having published several peer-reviewed papers myself (but these were on electronic properties...btw I can take measurements without "glass in the way" and I can analyse statistics for different samples.)

[TCO]

Originally posted by Tingkai
And I thought I was bad with my multiple posts.

Ramo has a habit of sliding from point to point. I am much more interested in having a "discussion" rather than a "debate". So I want to disaggregate the issues and label them clearly. If he makes a point that I agree with, I will say so.

Also, I have a hard time following posts longer than about a paragraph or so unless they are clearly organized into an essay. (buncha random cut and pastes doesn't have enough structure for me.)

[TCO]

Originally posted by Ramo

IQ studies have only a limited amout of credibility, even given optimal circumstances.

What's your beef? They use the same methods that people doing voter surveys or market research surveys use. They sample population in same way. Do you think all surveys are bogus?

You'd have to show me a real reason for a gap in intelligence between various peoples for me to believe it (or genetic evidence).

There are two seperate issues here. a. Wether an IQ gap exists. b. Wether this gap is geneticly caused.

Extensive testing (many different surveys, some with excellent sampling methods) has shown a 15 point gap in IQ between whites and blacks in the US. Do you contest this? Note that to see a gap does not mean that the gap is geneticly caused.

Sociology ain't physics.

What point are you trying to make?

[KrazyHorse]

Did "The Bell Curve" include a study of average IQ of Amerindians?

[TCO]

Originally posted by Ramo
But there's a clear reason why different amounts of melanin would be selected in different climates, while the reason for differences in the supposed smart gene is nonintuitive in the extreme. As far as I can see, if a few simple switches can turn you from stupid to smart, I see no good reason why the "stupid" alleles should be dominant in certain populations.

Real differences (i.e. not statistical anomalies) in intelligence arise from trade-offs. For instance, if I (speaking as a species) want to fly, I have to devote more energy to various muscles. So I either have to start eating more or lose my intelligence.

There's no comparable trade-off in the case of human peoples, so a difference likely does not exist.

1. Do you think all humans are equally intelligent? Would you agree that intelligence appears to run in families? Why aren't all families equally intelligent? Have you seen the twin studies that show a genetic component to IQ? (Note just because there is a genetic component to IQ doesn't mean race difference need be genetic in origin.)

2. Just because you don't understand the evolutionary history for a development or because you can't hypothesize a mechanism, doesn't allow you to ignore evidence. Most phenomena are identified experimentally first...than explained later. That's how science works...unless you are an Aristotelian.

BTW, you are falling to a circular fallacy here. A mechanism you don't like is proposed based on certain evidence. You disallow the supporting evidence since "it must be wrong" since it doesn't follow from a mechanism that you agree with.

3. Why aren't elephants smarter than people? They have nothing stopping them, right? (It is the pressure to develop intelligence not the limiters that is the issue...)

[TCO]

Originally posted by KrazyHorse
Did "The Bell Curve" include a study of average IQ of Amerindians?

No original research (NLSY). References to earlier studies.

[KrazyHorse]

'Kay, but did it include reference to earlier studies of Amerindian pop?

[TCO]

Originally posted by Ramo




You can't seriously be comparing this sociobiological theory, or whatever they're calling it, to QT!

1. I'm making a point about how science is done. You observe correlations and than investigate the cause. (Unless you are a disciple of Aristotle.)

2. I'm not a supporter per se of sociobiology. I just have an issue with your decalarative statements based on no evidence. Only criticism of other people's evidence.

And the differences regarding intellectual differences in peoples may have been "noted," but that's generally due to incorrect data.

Ok. Point me to some better surveys that support your point of view and tell me how they were better done (larger samples, etc.)

[TCO]

Originally posted by Ramo

Secondly, again, this boils down to time constraints. It's plausible for significant differences to develop after millions of years, but not 30 or 40000, with substancial sharing of genetic information between the peoples.

1. No duh. You said this before. What is the rate? If you want to make this argument, you need some support.

2. How much bulk intermixing do you think there was. Mitochondrial DNA shows bulk divergence of populations (at least in female lines...mDNA is only transmitted by the mother.) mDNA has been used to study various migrations of Indians, etc.

But a very small amount of variation in DNA between peoples relative to within peoples gives less credence to a significant variation in intelligence between peoples relative to within peoples.

Does this also rule out physical differences? They can't be genetically correlated to race either?

I don't know much about genetic variability in dogs, but can you point out a comparable piece of information to the one I pointed out regarding humans?

1. Already posted this once and its from an article you linked me to!

"In effect, he [President Clinton] was implying that there are no meaningful differences between populations. That belief is wrong and dangerously so," wrote Jon Entine in the San Francisco Examiner. "We share 98.4 percent of our genes with chimpanzees, 95 percent with dogs, and 74 percent with microscopic roundworms. Only one chromosome determines if one is born male or female. There is no discernible difference in the DNA of a wolf and a Labrador retriever, yet their inbred behavioral differences are immense. Clearly, what's meaningful is which genes differ and how they are patterned, not the percent of genes. A tiny number of genes can translate into huge functional differences."

I can't give you an exact number, but I've been told by bio-types (smart folk) that the number is much longer than 40000 years.

I need more than that. I need a rate. What is the absolute fastest that IQ could be changed in a population? We can call that the physical limit. Than the question is how much pressure was there on the population.

But mutations don't happen all that rapidly. And the vast majority of mutations are either ineffective or disadvantageous. The latter is particularly true when you're dealing with something as complex as the brain. The mutated allele has to get along with all of the other genes, and that isn't a simple task. In fact, I would wager that humans that more rapidly develop mutations in the brain are selected out of the population - the opposite of punctuated equilibrium.

IQ variance in a population or intelligence variation in dogs is not likely to be mutation limited. It is probably a phenomenon of concentration of certain traits already in the population. (And not perfect concentration either...since a distribution remains.)

It strikes me as inconcievable that a mutation could develop such that an animal becomes a whole standard deviation smarter within a 40000 years without any discernible selective pressures in the picture.

More of your circular fallacy that I described before.