Renowned pychometrician Charles Murray has given us, in The
Inequality Taboo, a concise summary of the most current science on
group differences in IQ and other measures of capability. Most of it
is not surprising to anybody who has been following the actual science
rather than press accounts severely distorted by the demands of
political correctness.
There is some new information here, however, and perhaps the most
interesting bit is that turns out to be much less to the Flynn effect
than meets the eye. The Flynn effect is the long-term rise in average
IQ scores recorded since IQ began to be measured in the early 20th
century. Advocates of the view that IQ is unimportant or meaningless
have seized on the Flynn effect to argue that IQ is either (a) a
statistical artifact, or (b) almost entirely environmentally driven
(and thus can presumptively be increased by correct social
policy).
Murray’s news is that the Flynn effect is not being driven by a
rise in average g, the measure of general mental ability that accounts
for over 50% of variance in almost all kinds of mental aptitude tests.
Since Spearman discovered the ‘g’ statistic, almost all psychometricians
have accepted that IQ is interesting precisely because it is a good
approxmation of g. Thus, the Flynn effect is basically a mirage —
it’s taking place in the noise, not the signal.
I’m not entirely sure what this means yet, and I don’t believe
Murray or other psychometricians have gotten to the bottom of it
either. But at minimum, it’s very suggestive that IQ differences are
either genetic or driven by environmental factors over which we have
little control. Spearman’s g, in particular, is notoriously
intractable. It is highly heritable according to separated-twin
studies. And while there is good evidence that it can be lowered from
its ‘natural’ genetic level by unfavorable environment (such as poor
childhood nutrition), it apparently can’t be raised by a favorable
one.
Indeed, Murray reports in a footnote evidence from a study in
Denmark that the Flynn effect has leveled off since the early 1990s.
Thus, it may be that we have already maxed out the effects of wealth
and better nutrition on the both the g and non-g components of IQ that
we can manipulate.
I believe that any serious consideration of the relevance of IQ must consider Howard Gardner’s Theory of Multiply Intelligences. When it is realized that IQ measures what Gardner identifies as the logical/mathematical intelligence – primarily, and that IQ deals very little with any of the other intelligences of Gardnes’ theory– which range from the social, the musical, the linguistic, the visual/spacial, to the kinesthetic, then much of the politically correct need to downplay ethnic and class based differences in the logic/math intellegence simpley evaperates.
For me the PC considerations then seem childish. Is it hard to admit to yourself that different people have different gifts? Isn’t it unreasonable to think we’ll all be equal – in all categories – all of the time?
I said this better on Slashdot, so I’m going to repeat myself here:
Fun word here: “reification”. The idea that just because you name something, it exists. The assumption is that intelligence is a single, rankable, immutable thing. (‘Cause if it isn’t, you can’t say that one person is better than another, and suddenly it matters a whole lot less.)
See, various tests tend to correlate with one another. The idea of IQ is to have a subject take scads of these tests in the hope of measuring the quality that they all have in common.
But this “general factor” is nothing more than a convenient mathematical abstraction. We don’t know if it represents a single quality of intelligence that white people tend to have more of than black people.
It’s a very seductive idea, this plan of ranking everyone in the whole world on a big scale from Smartest to Dumbest. But here, let’s take another example:
People’s height and weight tend to correlate. Taller people tend to be heavier. But let’s take the product of height and weight to get a Size Product, or SP. Now we can rank everyone in the world by size from Biggest to Tiniest. The problem is that our ranking doesn’t really say anything—we’ve squished two factors into one, and have lost a great deal of information. We’ve manufactured a number that seems to reflect size, but it’s just an abstraction.
So, is intelligence a real single entity, or is it just an abstraction of multiple quantities? I don’t think the answer has been conclusively established, and until it has, attempts to locate people on the Great Chain of Being are highly, highly suspect.
Perhaps separation, condescension and resentment are precisely
what the designers intended. A divided populace is much more ‘tractable’ than a united one.
hanzie.
“The world need not be that way. Any university or employer that genuinely applied a single set of standards for hiring, firing, admitting, and promoting would find that performance across different groups really is distributed indistinguishably.”
Which set of “groups” is he talking about here, races? Earlier he says ‘g’ is different between races and then here… is he saying performance isn’t (he is talking about academics as well as the workplace so that would mean g doesn’t correlate with performance academically…)? Is he referring to races when he says “groups” or am I missing something?
Charles Alexander: The goal is not to have an equal number of each group admitted. It is to admit the most worthy individuals.
Dear Eric,
are you sure the average IQ has grown or at least stayed the same? Didn’t it drop dramatically? I mean if we look at the movies, the books, the music, the political election manipulation techniques, the computer games and whatever from 1970 to 2005, we see that quality have amazingly dropped everywhere, and everything became much more stupid and drool-proof. So I think IQ must have dropped.
And actually it is a fact that clever people are likely to have less children than the others, so natural selection now works against IQ.
What do you think?
David McCabe: I am aware of that and I agree with it. I didn’t say otherwise. I just asked a question about something I wasn’t clear about from the article; your statement seems sort of tangential.
“Renowned pychometrician”
Please get your facts straight: Murray is a political scientist and IMO not a very good one.
Charles Alexander: The idea is that if you hire people of equal IQ in the first place, then you’ll end up with a racial distribution that does not reflect the distribution among applicants, but performance among those selected will be the same regardless of race, because you’ve already corrected for the racial difference. In other words, if you randomly pluck a white person and a black person from the street, there’s a significant chance that the white person will be smarter than the black person, but that is not at all valid if, instead of the street, you make your random selection from the pool of Nobel Prize winners (for the non-political prizes, i.e. not Peace or Literature).
Some analysts say that the Flynn effect may still be operational, but the tests show otherwise because they are no longer testing the same population.
I call bullshit on the Inequality Taboo link. I haven’t read the Bell Curve, so I’m just going off the article, but if it is any indication he’s been called a pseudoscientist because he is one. He names a number of fields, such as composing, and states that there has been no significant female contributor. According to who? Any creative field is subject to taste. More egregiously, and I’m surprised this wouldn’t be obvious to ESR, is that he states women have made no original contributions to philosophy. Maybe someone should point Murray to The Fountainhead…
Joseph Garvin:
Forget philosophy. I would wait ten years, and then point him to Lisa Randall. Time and the new Large Hadron Collider will tell us whether she and her colleagues are onto something.
Yes – Dr. Randall and her colleagues are on to something: Boondoogle Science. No planning at all is being made for the threat of unleashing baby black holes and mini-universes right here on Earth. I hope I am wrong, but I fear the worst from the LCH and other outrageously dangerous examples of Boondoogle Science.
http://dw-links.blogspot.com/2005/11/boondoggle-science-will-next-surprise.html
This thread appears to be dormant, so apologies for posting so late, but I just ran across it.
I don’t understand the logic of the argument that, if the Flynn Effect is “not due to changes in g”, it can’t undermine the validity of IQ testing as a genuine measure of mental ability. Let me take a quick walk through this from the beginning to see if I understand what’s supposed to be going on:
If you give a large group of people a battery of mental tests, their scores will be positively correlated: people who do better than average on one test will generally do better than average on all the others. The correlation between scores on various tests are almost always greater than 50%. “g” is the term used to denote this positive correlation.
Now take two populations, one from 1950 and one from 2000, and compare their scores. The *raw* scores from 2000 are significantly better on average (about one standard deviation) than those from 1950. (The actual IQ scores from any given population are normalized to have a mean of 100 and a standard deviation of 15, so you have to compare raw scores to see changes over time.) This is the Flynn Effect.
Now Murray comes out and says the Flynn Effect “isn’t due to increasing g”. Irritatingly, he always expresses it this way in his article, so I have no idea what sort of features of the actual raw scores he’s using to justify this claim. But the only way I can make sense of it is: the raw scores from 2000 are higher than from 1950, but the degree of positive correlation between them (which is what “g” is) is about the same. This is supposed to mean that, whatever the Flynn Effect is due to, it isn’t increasing general mental ability–which means the effect doesn’t undermine the claim of IQ tests to be genuinely measuring general mental ability, instead of some artifact of the tests.
I don’t see how this conclusion follows. Consider a parallel example. I take a large group of people and make a battery of measurements of body size on all of them: height, waist size, neck size, chest size, head size, length of arm and leg bones, and so on. All of these measurements are highly positively correlated (probably a lot more so than for mental ability tests)–people who are above average on one measure tend to be above average on them all. I use the term “s”, for “size”, to denote this positive correlation.
Now I compare scores from two populations, one from the US in 1800 and one from the US in 2000. The raw scores from 2000 are significantly larger on average than those from 1800. But the “s” factor–the degree to which scores from a given person are positively correlated–is about the same. Does this justify me in saying that the average size of the population didn’t increase from 1800 to 2000? As far as I can see, this type of claim is what Murray’s case amounts to. Have I missed something?