The Bell Curve

[_EAllusion]

I don't know how many people here read this, but the NYT had an article published a few weeks ago on a study led by researchers at Stanford, Harvard and the Census Bureau involving downward pressure on black class mobility that I think got a lot of attention:

https://www.nytimes.com/interactive/201 ... k-men.html

The study it is about is a big deal; the infographics the NYT uses are great; the conclusions are disheartening. This will be a memorable story for some time to come. The level of detail in the data used helps sharpen the picture of what is going on.

The key finding of the study is that being a black man, but not a black woman, has a very large negative effect on class mobility at every level of socioeconomic status. Quoting from the article,

"The study, based on anonymous earnings and demographic data for virtually all Americans now in their late 30s, debunks a number of other widely held hypotheses about income inequality. Gaps persisted even when black and white boys grew up in families with the same income, similar family structures, similar education levels and even similar levels of accumulated wealth.

The disparities that remain also can’t be explained by differences in cognitive ability, an argument made by people who cite racial gaps in test scores that appear for both black boys and girls. If such inherent differences existed by race, “you’ve got to explain to me why these putative ability differences aren’t handicapping women,” said David Grusky, a Stanford sociologist who has reviewed the research."

This is not to fault the Bell Curve for being published in 1994 on this point, but I did think it is relevant for a central argument that has come up on this page of this thread.

[_EAllusion]

I don't see how it makes sense to "increase an IQ" which is supposed to be innate (okay, what it's measuring is supposed to be innate).

I think we need to clarify here. So you have this notion of intelligence. There is disagreement among psychologists about what intelligence, if anything, is. Among psychologists, especially those who are interested in studying intelligence, the most common view is that factor analysis points to there being an underlying general intelligence ability called "g." g is mental ability that by increasing or decreasing it affects ability in a large range of mental tasks. Sam Harris was exaggerating a lot when he acted like nothing is on more sure footing in psychology. If you crack open an introductory psych text, you'll likely find a section on intelligence where multiple competing ideas are discussed. There are, in fact, 3 major theories of intelligence.

It's like saying that nothing is more established in philosophy than moral realism. Slow your roll, buddy. But it is true that this is a common view.

Ok, now that we have g, we also have disagreements about how valid IQ tests are for measuring g. The work Analytics is discussing says they are super-duper valid, but this is challengeable. Finally, not everyone who thinks g is real and IQ tests are a valid measure of it think g is fixed. The relative immutability of g close to birth is a central point in the Bell Curve, but it is subject to dispute. Plenty of people think intelligence is not analogous to the size of a cup you can fill up with aptitude, but rather the cup itself can expand and contract based on experience.

As you go down the line, each of these positions gets more and more challengeable.

I think Physics Guy's point is that opinions on intelligence among intelligence researchers might be like opinions on Marxism among academics who specialize in Marxist theory. (Or the LDS faith among those who would be getting degrees in Mormon studies.) The field just attracts people inclined to believe in the object of study. I'm sympathetic to the idea that this might be going on and would recommend at looking at what psychologists outside of the sub-sub-specialty of IQ testing think about IQ testing, but at the same time, the Bell Curve's views on it shouldn't be taken for granted. Again, there is no shortage of academic research out there challenging its core arguments.

Finally, I think you need to be careful about lumping all IQ tests together. Because the Bell Curve argues and then takes it for granted that the AFQT is just as good at measuring IQ as the WAIS, that doesn't necessarily mean this is so.

[_EAllusion]

IQ researchers may have published a lot of studies but this does't necessarily mean very much. Flat earthers have published studies. One of the basic problems with peer review is that peers are all people working in the same field. So peer review is fairly good at assessing quality of work according to the collective standards of the field but it's lousy at assessing the quality of the field as a whole. In my view it's quite plausible that IQ testing is really a pseudo-science based on mis-applying statistical criteria in cases where the assumptions on which the criteria are based do not hold.

That's not an expert judgement on my part. I don't have enough time or interest to get into the question deeply enough to be sure whether it's really all crap or not. Really confirming that kind of thing is quite hard because pseudo-fields may be hollow but it can take a long time to wade through all the fluff of obfuscation with which enthusiastic proponents have deluded themselves for a generation or more. Life is too short for me to decide I'm going to take down IQ—or satisfy myself that it's serious after all.

I'm still skeptical because the people who should be convincing me that IQ is real, people like Murray and Herrnstein, have done such a bad job. Their statements have raised obvious basic questions for me, which they themselves don't seem even to have noticed. Obvious basic questions sometimes do have good answers, but then competent workers in serious fields hurry to point out the questions and deliver their answers; they don't evade or ignore them.

One basic question concerns the innateness of IQ—its supposed un-trainability. In fact one can improve one's IQ score significantly by practicing the specific kinds of questions that appear on the tests. The only reason that IQ can be said not to improve with education in that way is that people just don't practice for IQ tests, because they would really have to go out of their way to do that. The questions on IQ tests are weird challenges that never come up in normal life. If there were any cross-training effect on IQ tests from real-world intellectual tasks, the claim that IQ is unaffected by education could never be made. Given this lack of influence of real-world intellectual skills on IQ, the converse claim that IQ nonetheless contributes generally to all intellectual tasks is eyebrow-raising.

Another basic pronblem for me is that reification is a much subtler issue than the IQ literature I've read seems to realize. As an analogy for what they're saying about intelligence, consider wealth. Some people are rich and some are poor, and it's arguably possible to determine a net financial worth for a person in order to quantify just how rich or poor they are. In that sense okay, wealth is a thing. And it's an important thing precisely because it is a general thing. One's ability to buy a yacht, or an election, doesn't depend much on what particular form one's wealth takes. The bottom line net worth is what matters.

And if you hear someone say all that about wealth, then so far, so good. Wealth is like that. If someone starts talking about the difficult issues involved in carrying your wealth around, though, or in storing it, then you have to do a double take. Do they imagine that wealth is gold coins? Bales of hay? Just because wealth is real, in a sense, doesn't mean that it's real in the way that gold coins or hay are. In some ways it's a thing but in other ways not, and if you casually treat it as a thing in one way just because it's a thing in the other, you're going to have a bad time.

One question I had for Analytics earlier on that I don't think was answered is if he thinks that IQ cannot be affected by educational background much, then are study aids that promise improvement on the ASVAB a scam? Hell, is studying for math tests in general a scam?

To one of your other points, this is one of the areas where the Flynn effect matters. What's been happening over time is IQ scores on IQ tests have been improving across the board quite rapidly - far more rapidly then an underlying biological change is likely to explain. It's over 5 points per decade. This is masked in reported scores because the tests have to keep being renormalized to take this into account. Go far enough back and a borderline score today would be an above average score back then.

Now think about Murray's view that job training programs as part of welfare are near useless.

http://www.pbs.org/thinktank/transcript129.html

All right, what are your and Dick Herrnstein'spolicy implications for all of this?

MR. MURRAY: Point number one is that the book's purpose is not asa setup for a five-point plan. 'The Bell Curve' is written to bringto a general audience some really important issues, and the policyrecommendations are secondary to that.

Having said that, there are a bunch of very specific kinds ofthings that the book points to. I'll give you an example. You want tohave a job training program for welfare mothers? You think that'sgoing to cure the welfare problem? Well, when you construct that jobtraining program and try to decide what jobs they might qualify for,you had better keep in mind that the mean IQ of welfare mothers issomewhere in the '80s, which means that you have certain limitationsin what you're going to accomplish.

Do you mean to tell me that we can't construct a job training program that would be helpful in the economy when this interview was conducted for a person who would've scored average decades ago? How limited, exactly, was a person with an IQ of 105 in 1950? Here, the implication is that borderline IQ scores reflect borderline IQ ability. Otherwise, Murray's comment makes no sense. But if we assume IQ tests have been consistently valid and real intelligence is improving rapidly due to the Flynn effect, then how much intelligence do you really need to have job training to obtain good, gainful employment?

If real intelligence hasn't been improving and this only reflects improved testing ability, doesn't this speak to that issue instead, then? It's a catch-22 in this context.

[_Analytics]

IQ, as measured by the AFQT, is your ability to answer questions about Algebra (among other things). It seems that you don't get this basic point. It turns out, according to Murray, that how good your Algebra class is relatively speaking actually has very little to do with how good you are at Algebra relative to others.

I don't consider people who don't understand the concept of statistical correlation to be in a very good position to evaluate what Murray thinks about such things.

(P.S. Studying for IQ tests, even the kind you are referring to, does actually move your IQ.)

I'm glad you and Murray agree about something.

[_Doctor CamNC4Me]

Wouldn't studying for a test in of itself be an indicator of one's IQ?

- Doc

[_EAllusion]

I don't consider people who don't understand the concept of statistical correlation to be in a very good position to evaluate what Murray thinks about such things.

Lol. First, the correlation between AFQT and IQ tests more focused on abstract mental tasks isn't high enough to deserve this response even if it didn't miss the point. Second, "They are not intended to test how well you learned any given school curriculum. You don't seem to grasp this basic point." is a bad line of thought in this context when we are talking about an aptitude test that is proxy converted into a correlated IQ equivalent. The same can, and is, done with SAT scores. Since that is our measure of IQ and we are using that data set to say that variance in quality of schooling doesn't affect that measurement of IQ, then we are saying that variance in quality of schooling doesn't affect those tested aptitudes. If it does, then the line of argument collapses.

I'm not sure this point would be lost on the esteemed Dr. Murray. This is a feature, not a bug in the line of argument. Murray thinks that there's a small cognitive elite that should be given high end academic resources, and the money is poorly spent on the rest. They, mostly disadvantaged social classes, should be directed away from college and into job training programs that focus on developing skills more appropriate to their native ability rather than spending money squeezing performance on math problems and reading comprehension out of what doesn't produce much when squeezed.

College for disproportionately well-to-do whites and working class vocational training for mostly poor minorities. It's the natural order of things.

[_EAllusion]

Wouldn't studying for a test in of itself be an indicator of one's IQ?

- Doc

That would suggest some conscientiousness or a support system around a person directing them to behave in a conscientious fashion, but it wouldn't tell you much about that person's general mental ability.

[_Analytics]

Hi Analytics,

It sounds like you might not get back to this soon. That said, perhaps I wasn't making my question clear as to what I was asking.

As noted throughout this thread, no one seems to be arguing that the value of education is tied to improving IQ. Nor should anyone expect that raising IQ would be the target of education. We've also discussed that there are varied views as to how much IQ really measures general intelligence and how much it represents ability to perform to a standard defined by a particular cultural background. We can say that Murray favors viewing IQ as measuring something more universal but that's merely noting his position. Whether one is skeptical of IQ or favorable, no one would be surprised to see high IQ's correspond with high levels of success in western cultures, in particular in the US. Like the cartoon, it's like someone saying monkeys perform very well on average compared to other species when it comes to climbing trees. There's something almost tautological to this type of claim.

So, again, what's the actual argument here that matters?

Let me try framing it this way. America has lots of problems and challenges, including growing inequality, poverty, crime, educational disparity, health issues, etc. As we try to deal with these issues, we need to understand the sociological landscape, and we need ways to model the nature of the problems so that we can figure out solutions.

According to mainstream thought, the problems fall along racial lines. People of an historically unfavored race are failing much too often. Therefore, the solution must be to deal with the racism that must have caused all of this. Merely ending most of the racism has helped, but huge problems remain. Thus, the solution must involve affirmative action and such things. Since blacks were systematically disadvantaged for generations, as a society we must systematically give that race an advantage. Hopefully after several generations of affirmative action and whatever else we can think of to erase the damage done to the race because of past and lingering racism, the underperforming race will eventually perform as well as other races. Once the systematic problems of racism are dealt with, members of that race will be able to reach their full potential and do everything whites can do--even become an astrophysicist or the President.

That is the current, mainstream paradigm, more or less.

Murray is suggesting that blacks with low IQs have a lot more in common with whites of low IQs than they do with other blacks. Likewise, blacks with high IQs have more in common with whites with high IQs than with other blacks. In fact, most of the problems associated with race issues can be systematically explained as really being about IQ, irrespective of race. This raises the question about the real nature of these issues.

This all isn't to say that racism and prejudice don't exist, and it certainly isn't to say that racism is somehow justified. Rather, it suggests that we should try to understand what is really driving These problems. If the issues are really about IQ, that probably needs to be understood to find the best solutions.

So the objective is to figure out how to do a better job of raising the IQ of everybody who is dull. Better nutrition, better environments, and better education are part of it, as is research to figure out what else can be done or how it could be done better. As we strive to maximize everyone's IQ, which can be thought of as maximizing their potential, we need to work on helping them reach their potential.

As I read the book, that is his point--IQ matters. The only reason he brings up race is to point out that some problems associated with race might really be about IQ instead.

[_Analytics]

You're aware the AFQT score is compiled from several of the subsections of the ASVAB and this is later proxy converted to an IQ equivalent, right? Murray and Hernstein regard this as a great IQ measure and it informs their work throughout.

Those subsections test academic knowledge, so scores are necessarily related to one's academic knowledge.

When I was a Mormon missionary, i got into an extended argument with a guy about whether Mormons believed in Jesus. Regardless of what I said, I could not convince him Mormons were Christians. He was positive he knew everything there was to know about the Mormon Church, and he had no interest in listening to my point of view.

After a while, I saw no point in continuing the conversation.

[_EAllusion]

The only reason he brings up race is to point out that some problems associated with race might really be about IQ instead.

This is absolutely disingenuous. They have an entire section concerned with the "dysgenic" pressure elevated black (and latino) fertility rates have on the US population and within minority communities. This is jargon from race science / eugenics crowd that refers to later generations inheriting undesireable traits, intelligence in this case, from the previous generations. They talk about as a real concern in general and in the context of policy, like when they point out a concern that welfare programs encourage fertility among low IQ minorities.

Open up to page 352, read that section, and explain how race is only brought up to point out that problems associated with race might really be about IQ. ("Might" is also a dubious underselling of what actually is being argued.)

Or, to pick another example, they spend some time arguing that Affirmative Action is driving down the quality of the workforce by forcing employers to hire lower IQ minorities into positions over higher IQ whites. This is used to make policy recommendations to eliminate or ease affirmative action programs and replace them with alternative approaches. They give several options for what those alternative approaches might look like and in the end ultimately reject them. The finish with appealing to the reader to strongly consider ending racial anti-discrimination laws in general. This is, in fact, about race and IQ.