Bill Snedden

There isn't really a "control" in the sense of "experimental control" in the type of analysis conducted by Hernnstein and Murray. They used data gathered from multiple sources, but mostly from the National Longitudinal Study of Youth. The data is gathered as the results of tests and self-reported/proctor-indicated demographic information.

However, the authors conducted a factor analysis in which they did seek to "factor out" the impacts of educational level, so that their results would not be affected by it.

Regards,

Bill Snedden

Bill Snedden

But that's my point. While skin color is certainly the result of a genetic trait, it hasn't been demonstrated that the genes coding for skin color belong to some group of genes that can be identified as belonging to any particular group of individuals that we might label "african americans" and no other persons.

There are people who identify themselves as "african-american" whose skin is as light as mine. Likewise there are "caucasians" with rather dark skin.

Regards,

Bill Snedden

Bill Snedden

Yes, I've heard that as well. However, regardless of intra-disciplinary acceptability, the fact remains that weak correlations yield weak predictive value. I wouldn't say that a 30% probability provides any type of realistic prediction upon which to base a decision. If it's the best we have to go on (which is apparently often true in the social science arena), then we must make do but at the same time we must realize that we are more likely to be wrong than right!

Regards,

Bill Snedden

ohwilleke

I don't think that the Bell Curve is a great book, but I am also not impressed with some of the criticisms targetted at it.

Race exists

Some critics of the Bell Curve attack it because they claim that race doesn't exist biologically.

The biological non-existence of race argument is one I would quibble with. In the United States, in the later part of the 20th century, which is the time period that the book in question is addressing, race is a quite well defined sociological reality as applied to the vast majority (certainly at least 90%-95%) of the population. A social or physical scientists who claims to be unable to distinguish between a white American and a black American in the vast majority of cases, has his head in the sand, and is failing to observe the obvious, which is bad science.

Moreover, in the United States, in the later part of the 20th century, the racial distinctions made by a layman who is noticing this sociological reality have a very strong (if not absolutely perfect) connection to a person's ancestry.

Yes, race in the United States is more complex than a simple black and white division. Yes, those people who are part of the sociologically defined group that a layman understands to be "black" includes both the smaller number of people who are "pure blooded" African in ancestry, and the larger number of people who have mixed African and white blood. Yes, the rubric Hispanic encompasses a variety of people who look different from each other, and are clearly neither black nor white. And yes, Native Americans, Asian Americans (several varieties), Arab Americans, and many people who are of mixed race, also do not fall clearly into either the black nor the white category.

But, to proclaim that race does not exist, when race is a very real part of the reality all of us deal with from day to day, and when this reality is closely (if not exactly) tied to a person's ancestry, is to be oblivious to what is going on around you.

And, I am skeptical that problems in defining race in a meaningful way, are an important flaw in the Bell Curve's analysis. Flaws in defining race may have a marginal effect, but I have a hard time seeing how this particular problem could change the thrust of the analysis.

Intelligence exists.

Some critics of the Bell Curve argue that intelligence isn't a meaningful concept.

I disagree. Intelligence tests are good predictors of academic success, economic success within the "normal range" or less (intelligence is not a great predictor of who will be superrich or superhigh income), and in the above average to low range, a person's likelihood of experiencing serious economic or social problems (unemployment, poverty, out of wedlock teen pregnancy, divorce, conviction of a crime, less than honorable military discharge, etc.). Intelligence is not a very good predictor of who among people with above average or better scores on intelligence tests will experience serious economic or social problems.

These tests aren't perfect predictors, but these tests are better predictors than race, education and parental socio-economic success, even when combined. What precisely these tests is measuring isn't perfectly defined, but a wide variety of tests of ability, from military aptititude tests, to SAT general tests, to GRE tests, to vocabulary tests, to tests that call themselves intelligence tests corrolate very strongly with each other, and have similar predictive power.

Does an intelligence score tell you everything there is to know about a person? No. Does it tell you something that it important in a lot of contemporary social and economic situations? Yes.

Intelligence has a significant hereditary component.

Some critics of the Bell Curve doubt that intelligence has a significant hereditary component.

A variety of studies have examined how much of what intelligence tests measure is hereditary. Some of those studies attribute as little as 30% of the variablity observed in intelligence to nature, and 70% to nuture. Some of those studies attribute as much as 80% of the variablity observed in adult intelligence to nature, and a mere 20% to nuture. No serious studies find that there is no significant hereditary component to intelligence.

Those who has seriously studied the issue agree that there is a significant nature component to intelligence. These studies have looked at twins (separated and not), at adopted children, at siblings (separated and not), at parents and children (separated and not), and just about every other conceivable research design. To find a nature component of intelligence variability as low as 30% is out of step with most of the research. To find a nature component as high as 80% of the variability is beyond what most of the research shows. Most of the studies estimate that 40%-60% of the variability observed in intelligence is due to nature, which, for a social science, is certainly a significant figure.

Moreover, even studies that find a middle to low amount of variability due to nature in intelligence, at least find that a significant part of the nuture component of intelligence has its source early in life. For example, a number of studies have found that alcohol consumption during pregnancy, whether or not a child was born prematurely, and the number of months of breastfeeding a child receives in the first year of life, have significant and measurable effects on adult intelligence.

By a child's teen or even pre-teen years, and certainly by early adulthood, that child's intelligence as measured by intelligence tests will be very stable for the duration of that person's life. For a very large proportion of children (although there are important excepts which are very relevant to the issue of how race and intelligence are connected), the same parents who are the source of what has been isolated as a hereditary component of intelligence are also the driving influence of much of the nuture component of intelligence.

Thus, variability in intelligence has a very signficant hereditary component, and who your parents are, assuming that you do grow up with your parents as most but not all people do, is an overwhelming influence on the variability observed in children's intelligence.

Intelligence is distributed in something like a Bell Curve

Intelligence tests really do show that there are lots of people who get similar scores on these tests which are neither exceptionally high, nor exceptionally low. People who score exceptionally highly on these tests are rare. People who score exceptionally poorly on these tests are rare. And, as noted above, these tests really do reflect, at least in a gross fashion, how the people who take them fair in the real world.

Admittedly, statisticians have designed the scoring of the tests to fit a Bell Curve (i.e. "normal" distribution). But, the evidence does suggest that the basic scheme in which there are lots of people in the middle of the intelligence distribution, and proportionately far fewer people at either extreme, is correct. And, there is reasonable evidence that this same general pattern of test performance is found among subgroups of the total population based on race and class.

The Flaws

1. The degree to which the bulk of studies find that intelligence is hereditary is not great enough to rule out the hypothesis that racial differences are due to nuture rather than nature components of the variability in intelligence.

This is very important because, if nuture is really at fault in racial differences (presumably due to historical causes rooted in slavery for which the United States as a society bears some resposibility), then affirmative action in social services and education makes policy sense. If nature is really the source of racial differences, then expecting equality of results in a uptopian future is unrealistic in a meritocratic society, and one has to question if holding individuals responsible for outcomes strongly linked to their intelligence as a meritocratic society does is moral, since those who perform poorly couldn't improve their abilities if they wanted to do so.

2. The Bell Curve's analysis does a poor job of explaining why intelligence scorces have systematically risen over long periods of time; a reality that argues against the importance of inheritance in intelligence.

This also goes to the nature v. nuture, which, as noted above, is very important for policy reasons, yet hard to establish empirically.

3. The Bell Curve's analysis does a poor job of showing whether intelligence is really a measure of culturally neutral ability (although many attempts have been made to show that this is in fact true), or a measure of ability to succeed in a particular culture. In other words, it doesn't show us if a particular black person's low score on an intelligence test reflects an overall lack of intellectual capacity regardless of context, or just a lack of ability to succeed culturally and economically within the context of a society run by predominantly white elites who have very different cultural norms than the society that this particular black person grew up in.

For many policy purposes, it really doesn't matter which version of this dispute is true, since it is hard to change society. But, an argument that intelligence is a culturally dependent matter, does support arguments that "separate but equal" may not have been such a bad idea, and that "Afrocentrism" is an important concept which should be given a place in our society.

4. The Bell Curve does a poor job of explaining why anyone should care about average group performance on intelligence tests, as opposed to individual performance on intelligence tests.

We already live in a society that strongly distrusts racial stereotypes in favor of treating people like individuals. Few of us want to turn back that clock because we has history to show us the harm that racial stereotyping can cause. The Bell Curve admits that differences between members of groups often exceed the differences between groups. Hence, it is hard to see what a group based analysis based on race adds to policy decision making. Arguably it just adds mud to issues that shouldn't involve race.

5. The Bell Curve draws policy conclusions that don't follow closely from the research that is used to support those policy conclusions.

Most reviewers agree that the policy conclusions seem tacked on and ill considered. The nature of intelligence, and in particular, the degree to which it is hereditary and connected to racial identity are important to some important policy considerations, but the author's conclusions appear to be knee jerk conservative ones that force a reader to examine the rest of the statements made in the book for conservative bias, rather than thoughtful and careful implications of the research itself. This is particularly problematic given the author's association with the conservative heritage foundation.

[ May 15, 2002: Message edited by: ohwilleke ]

[ May 15, 2002: Message edited by: ohwilleke ]</p>

seebs

<a href="http://www.lrainc.com/swtaboo/stalkers/jpr_gould_paid.html" target="_blank">http://www.lrainc.com/swtaboo/stalkers/jpr_gould_paid.html</a>

(I hope that shows up.)

I don't have a strong opinion either way, but I think it's pretty clear that Gould (mayherestinpeace) was not at his best in his attacks on The Bell Curve.

Is this real? I don't know. I know that other physical characteristics appear to vary with racial groups. I have no underlying reason to believe that, unique among all the things we can measure in people, this one *wouldn't* vary with ethnic background.

I also see that this has the potential to destroy modern civilization, because we've built so much on the assumptions that:

1. Intelligence *defines* your value to the world.
2. All groups should be treated equally.

I have no solution. The whole thing bugs me, and I don't know how to resolve it.

Daydreamer

1. I agree that the evidence seems more correlational than causational.

2. IQ Tests - even the new ones - only measure three sub-intellegences: Math/logic, spatial, and linquistic. Last I heard there were five more: interpersonal, intrapersonal, musical, and two others.

3. This fails to take into account non-intellegence factors, such as personality etc. I remember vaguely hearing of an interesting parallel between mild Autism and famous cryptographers/mathamaticians, like the one featured in A Beautiful Mind, somewhere...