The Political Economy of IQ, Or Tilting At Windmills with Steve Hsu (and Jason Richwine)

Steve Hsu has been on a tear lately. Giving talks about IQ, here and , and partnering with BGI to sequence the genomes of “high cognition” individuals in a quest to solve the giant “problem” of IQ. This effort has hit Vice magazine, Slate, and, more recently, NPR. To give you the CliffsNotes version of Hsu’s argument: IQ is a quantitative trait, just like height and can therefore be selected for improvement.

If you don’t know Hsu, he is the vice president for graduate admissions and research at Michigan State University. He has input into the direction of MSU’s research efforts and whom MSU admits to its graduate programs. .

Obviously, Hsu isn’t the first to claim a method of ranking humans according to cognitive endowment, nor will he be the last. The first attempt was Spearman’s 1904 paper in which he noted a positive correlation in schoolchildren across seemingly unrelated subjects. He called this construction “General Intelligence.” Hsu’s attempt is one of the latest and in keeping with intellectual fashion, he has substituted “Genetic Architecture” and a series of quantitative trait loci for Spearman’s construction, but otherwise their respective projects follow the same intellectual current.

I will raise three lines of objection to Hsu’s claim to have found the “genetic architecture” of high IQ. The first objection is that the tests Hsu uses to determine IQ are culture bound. That is, they reward some types of test takers more than others. Second, Hsu’s model assumes that certain types of knowledge require more cognitive ability to master than other forms do. Third, Hsu falls victim to the utilitarian fallacy. I’ll briefly discuss the first two before turning to the third.

First, a well-worn criticism of Hsu’s work on the genetics of “higher cognition” is that the SAT, which he uses as a proxy for the g factor (see Gould for a thorough debunking) is slanted towards rewarding wealthy test takers. This is a well understood phenomena. The more money one’s parents make (or the more academic their jobs are!), the better one’s chances of succeeding on the SAT.

Hsu, for his part, has acknowledged the culture bound nature of his quest, but seems to believe he can get around the unfortunate fact of the SAT with a “culture-neutral” test, though he does not actually have any concrete plans to administer it and his work uses the SAT extensively. Unfortunately for Hsu, people who think in “culturally neutral” categories are about as common as Giants on the plains of La Mancha.

 

Second, in this paper, Hsu attempts to link certain majors (physics and mathematics) to “cognitive thresholds.” In doing so, he assumes a natural order of cognition running from fields requiring advanced mathematics to those do not. There is no need to do too much here work tying this up with Victorian conceptions of the savage and the civilized. Suffice to say there is a long body of work within anthropology (and cross-cultural psychology) which explodes this assumption. What is important is context. Some skills are more important, and hence likely to receive more development in certain contexts as opposed to others. Hsu is simply mistaking an epiphenomena of formal schooling in the early 21st century for an eternal human truth.

For example, in 1904, Spearman concluded that the most powerful correlation for “General Intelligence” was excellence in Classics, followed by Common Sense, Pitch Discrimination and French. Mathematics is on the list and and correlated positively, but the highest correlations were dominated by language. It is unsurprising that among English schoolchildren in the first half of the 20th century, language and music would be highly valued within formal schooling while among American college applicants in the early 21st century, mathematical ability would be highly valued.

In each case, the sample subjects are cultivating historically and socially contingent subjects taught within a powerfully hegemonic institution. What makes Spearman and Hsu irresponsible social scientists is i), their unreflective assumptions of these categories as eternal and unchanging and ii), their use of these categories to rank and sort humans in a manner that closes off their possible futures. In this way, both hew closely to deficit models of school failure.

Third, the utilitarian gambit is a staple of quantitative approaches to human action, and as with any mathematical model, must make some assumption about motivation as a start. Hsu simply assumes everyone who takes the test tries to get the best score. He uses the brute fact of admissions based on the SAT to model a market for IQ, which sorts test takers according to their potential with the underlying assumption that test takers do their best to gain high scores. Thus, a political economy is formed whereby high scoring test takers are slotted into the most competitive majors (Hsu thinks some majors require more cognitive ability than others) at the most competitive universities.

However, we know from a wide and deep body of literature within the Anthropology of Education that not all test takers apply themselves equally or are motivated to do well on tests they know are of little use to them. Hsu knows this as well, but he chooses to ignore it with his work on genetics and IQ. However, in an interview with Psychology Today, Hsu gave the following answer to a question about Richard Feyman’s allegedly low IQ score:

3.  Is it true Feynman’s IQ score was only 125?

Feynman was universally regarded as one of the fastest thinking and most creative theorists in his generation. Yet it has been reported-including by Feynman himself-that he only obtained a score of 125 on a school IQ test. I suspect that this test emphasized verbal, as opposed to mathematical, ability. Feynman received the highest score in the country by a large margin on the notoriously difficult Putnam mathematics competition exam, although he joined the MIT team on short notice and did not prepare for the test. He also reportedly had the highest scores on record on the math/physics graduate admission exams at Princeton. It seems quite possible to me that Feynman’s cognitive abilities might have been a bit lopsided-his vocabulary and verbal ability were well above average, but perhaps not as great as his mathematical abilities. I recall looking at excerpts from a notebook Feynman kept while an undergraduate. While the notes covered very advanced topics for an undergraduate-including general relativity and the Dirac equation-it also contained a number of misspellings and grammatical errors. I doubt Feynman cared very much about such things.

How the world has changed since 1904!

Hsu is more than eager to judge you based on your SAT, ACT, GRE,  IQ, or Wonderlic score, but for his personal hero, an excuse must be made. And the excuse he makes for Feynman exposes the assumptions underlying his theory. Feynman, of course, is gambling that he can afford not to care about some things. But, he isn’t alone. Everyone gambles on what they can and cannot afford to care about.

One of Hsu’s examples of an acceptable SAT substitute is the Wonderlic. The Wonderlic is perhaps most famous for its use in the NFL scouting process. Morris Claiborne, a cornerback from LSU, infamously scored a 6 on the Wonderlic. What does he say about the IQ test?

“That test don’t tell me who I am and what time of guy I am and what kind of ability I have. That test can’t drop me.

“They say it’s an IQ test. I came to the combine for football. I looked at the test, and wasn’t any questions about football. I didn’t see no point in the test. I’m not in school anymore. I didn’t complete it. I only finished 15 or 18 questions.”

Claiborne was selected number 6 in the NFL draft and signed to a contract worth several million dollars. Obviously, Claiborne plays football and does not teach theoretical physics, but his comments are illuminating for a powerful reason.

Claiborne is just like Feynman in that he has the good sense to know what not to care about, and where to focus his attention. In fact, the logical justification Claiborne uses in blowing off the test and the logical justification Hsu uses in explaining away Feynman’s relatively low IQ score are almost identical.

Hsu on Feynman:

I doubt Feynman cared very much about such things.

Claiborne on Claiborne:

I’m not in school anymore. I didn’t complete it.

What Feynman and Claiborne are doing is deeply playing with the boundaries of their respective fields. As Geertz observed in a Balinese cockpit deep play is two things. It is 1) instructive work and 2) metasocial commentary. That is, it is a set of instructions about how to do the thing we are now doing and what the consequences might be (or might not be) for doing it wrong.

Though these two examples are playful in the sense that the stakes are low and nothing much is to be lost or gained by a CB who refuses formal logic or a physicist who takes an eccentric approach to spelling, they both point to the metacognitive nature of intelligence and the difficulty of pinning down something like the “g-factor” without recourse to the messy details of context or taking into account the inevitability of self-reflection. Such as: Why am I writing this post?

Postscript:

I sat on this post for a long time. I thought perhaps it would be better to take another direction. But this morning’s news carried the tale of Jason Richwine and a reminder of why I might write about IQ and not something else. Richwine is the co-author of the Heritage Foundation’s new report on immigration and the author of a Harvard dissertation titled “IQ and Immigration Policy.” Here is the abstract:

The statistical construct known as IQ can reliably estimate general mental ability, or intelligence. The average IQ of immigrants in the United States is substantially lower than that of the white native population, and the difference is likely to persist over several generations. The consequences are a lack of socioeconomic assimilation among low-IQ immigrant groups, more underclass behavior, less social trust, and an increase in the proportion of unskilled workers in the American labor market. Selecting high-IQ immigrants would ameliorate these problems in the U.S., while at the same time benefiting smart potential immigrants who lack educational access in their home countries.

In a few days, I will set Richwine (and Hsu) in the context of the Culture of Poverty debates. Suffice to say, they offer nothing new to debates over IQ, or poverty or immigration. Their innovation lies in the naked, unreflective application of a naïve sociobiology to policy debates over access to democratic institutions – citizenship and public education.

Update:

Richwine is out at the Heritage Foundation. Meanwhile, Richwine’s advisor has been distancing himself at a rapid clip:

“I have never worked on anything even remotely related to IQ, so don’t really know what to think about the relation between IQ, immigration, etc,” Borjas told me in an email. “In fact, as I know I told Jason early on since I’ve long believed this, I don’t find the IQ academic work all that interesting. Economic outcomes and IQ are only weakly related, and IQ only measures one kind of ability. I’ve been lucky to have met many high-IQ people in academia who are total losers, and many smart, but not super-smart people, who are incredibly successful because of persistence, motivation, etc. So I just think that, on the whole, the focus on IQ is a bit misguided.”

62 thoughts on “The Political Economy of IQ, Or Tilting At Windmills with Steve Hsu (and Jason Richwine)

  1. Michael,
    Nice blog. It is interesting how the belief in intelligence=i.q.=generla excellence formula keeps popping up in modern cultures, with the net effect of legitimating pre-existing systems of inequality. I really like Gould’s book “The Mismeasure of Man,” while still acknowledging the critiques that have emerged since its publication. In my view, Gould’s underlying argument about the use of biological sciences to legitimate pre-existing systems of ethnic and racial stratification remains unchallenged.

    Tony

  2. It is really a naive form of functionalism. From Comte forward, biology has always been used as an analogue for social organization. Or, maybe it would be better to say from Plato’s Republic forward. Essentially, this is what Gould is saying about the use and abuse of biology in the social realm.

    In this formula, biological adaptation equals functional consensus.

    I think the real use of Gould is that he points out that adaptation in evolutionary theory has limits and there are other modes worth thinking about. Biology may yet have some useful metaphors.

  3. LOL @ the Gould reference. Are you going to cite refutations of phrenology as reason to throw out neuroscience next?

  4. I’m going to remember this is ten years. If Hsu succeeds in is endeavors you will still deny he’s right. That’s pretty much the worst kind of person i can imagine.

  5. @Leon. I don’t think that Michael is asking anyone to throw out biological sciences, though he is asking biologists to think about how culture effects culture, too.

    Phrenology as a response to neuroscience is probably a bad idea. But, I do think that a good case can be made that neuroscientists studying mirror neurons might profitably spend some time reading about the nature of the “looking glass self” developed by Charles Horton Cooley, George Herbert Mead, and their successors.

  6. I reckon it is time to fulfill the promise I made to Tony in the comments section of an earlier post. I had lots to say about Stephen Jay Gould and his charlatanry back then, and I suppose now is as good of time any to pick up where I left off.

    http://www.ethnography.com/2013/03/human-genetics-and-social-theories/comment-page-1/#comment-9905

    You assert that IQ tests are indeed culture-bound, and they are biased in favor of cognitive abilities that our modern, industrial civilization happens to privilege above any others. To that, I say — pardon my Gallic temper, but no SH*T, Sherlock. That’s precisely what they were designed to do. (Here, you and I have achieved a complete synchronicity of opinions, and I plan to savor this rare coincidence while it lasts.) And that’s the only thing that IQ tests need to do — that is, measure the only definition of intelligence that matters most in the era of microprocessors and computer programming.

    But of course, from here on, our opinions start to meander in largely orthogonal directions. You emphasize that other cultures honor definitions of intelligence that are largely different from our own, and to that, I largely assent — well of course they do. Nonetheless, I will beg to differ. If things like voodoo, traditional Chinese medicine, and magical potions actually worked (let’s add bloodletting and homeopathy to that list, lest I offend your readers with my naked Eurocentric bias), we might actually have to take them seriously. But that’s just too bad, because they don’t work at all. In the year 2013, there is absolutely no substitute for clean drinking water, antibiotics, planes and automobiles, electricity, engineering, neurosurgery, and modern medicine. (Guess which civilization made that work? Surely not the Chamorros or the Mbuti pygmies.) If the entire French literature department at Stanford U perished in a tragic shipwreck, the universe would be nary the wiser, yet we all would be forced to notice if the world lost a single Feynman, Gauss, Pasteur, or Salk. (Yes indeed, we would eventually notice to the tune of millions of dead children!) Certain cognitive abilities matter more than others according to Maslow’s hierarchy of needs. Whether you like it or not, modern infrastructure, electricity, abundant food, vaccines, a navigable legal system, and clean drinking water should come first in our lives. Everything else is window dressing. The fact that Steve Hsu and others like him prioritize STEM over the so-called “fluff” disciplines is not a mere reflection of their prejudice against the belles lettres — STEM pays for not only your ethnography, and my love for English literature, but everything else in between.

    And of course, lest I forget, our modern technology is also a source of intermittent misery for those who have no access modern conveniences — whether it be because of a poor facility with Western “ways of knowing”, or due to institutional disadvantages that are largely beyond their control. It also is directly, or indirectly responsible for global warming, the desecration of non-Western cultures, and the wanton environmental destruction we see in much of the world today. But that’s too bad, because you can only stop the evil consequences of science with even more science. We’ve already opened Pandora’s box, and there will be no worldwide reversion to a hunter-gatherer lifestyle any time soon. And if you care about your progeny, you’d better pray that such a historical scenario never occurs!

    Furthermore, this is not to assert that there aren’t academic disciplines that aren’t as cognitively challenging as physics or neuroscience. (Let me recall that during WWII, Ruth Benedict authored a book that was so prescient in its observations that even the Japanese today attest to its accuracy, all despite the fact that she never once stepped foot in the island nation. You can bet your ass Benedict would have scored well above the mean on the WISC.) Far be it the case. We even have preliminary evidence that classicists might be the smartest of all graduate students, even brighter, on average, than lowly physical scientists like Steve Hsu.

    Well, and truly, what a delightful coincidence that IQ tests do a decent (albeit imperfect) job predicting lifelong performance in each and every one of these disciplines; your cultural anthropology included. This is indeed the mathematical basis for “g” — whether or not intelligence actually exists as a linear construct that may be measured on a unidimensional scale is irrelevant. (And in either case, that last observation is a caricature, because IQ scores are a dimensionless quality that are measured by deviation from the mean. But I digress) The covariance matrix for correlations between IQ scores yields a single principal component that by far explains the vast majority of the variance in outcomes — it matters not whether the correlations exist purely by construction, or whether or not they merely reflect the existence of separate, orthogonal “intelligences” that are measured together in each element of the testing manifold. Statistical artifice or not, spearman’s “g” has predictive value, and that is all that counts when we measure “intelligence”.

    Which brings me to my next point. It is well known that the correlations between IQ scores, health, academic prowess, and social mobility are some of the highest ever recorded in the social sciences. And judging from longitudinal studies of high IQ children born into meager circumstances, the direction of causation is largely in one direction. Higher IQ predicts better performance in school, greater competence in one’s chosen field, a higher amount of income accrued throughout one’s lifetime, etc. — yes, even when controlling for economic privilege. The fact that children of the wealthy score higher on these tests is a given. But that hardly proves that IQ tests are biased, any more than the metric system is biased, because the wealthy tend to be taller than the poor. Cognitive tests have roughly the exact same predictive value for all people just about everywhere. Nobody who scores below 100 on an IQ test is going to become a neuroscientist anytime soon.

    We also know that IQ scores are heritable, and the correlation between two scores taken by the same person under different environmental circumstances (~0.9) is high enough that on a group level, we may regard “IQ” as a phenotype that remains more or less constant throughout the lifespan. (On a side note, ethologists measure animal behavior along much the same lines, yet nobody trashes them for it. Just how does one “objectively” measure the aggressiveness of a cape buffalo? Yet we know heritable differences must exist.) The twin and longitudinal adoption studies indicate that the heritability of IQ is roughly less than that of height, under a broad range of environments within developed countries. And if you have heard somewhere that twin studies are irredeemably flawed, racist, or evil because of their eugenicist origins (the typical culprits behind this smear campaign in academia are Gould, Kamin, and Lewontin), look no further, because we have GWAS studies that largely replicate the same figures. Knowing the neurological correlates of IQ with actual, measurable dimensions of the brain (e.g. brain size, volume of white matter, glucose metabolism rate, etc.), most of which range from 0.4 to 0.6, it becomes an inescapable conclusion that as crude as IQ tests are, they measure some property (or linear combination of properties) of the nervous system that corresponds to what we call “intelligence” in the context of modern civilization. Attempting to decipher the genetic architecture of an amorphous trait like “intelligence” is therefore hardly the wild goose chase that you portray it to be.

    Note that up to this point, I have not said anything whatsoever about the immutability of IQ differences between human populations. We know they exist, and Richwine should never have been rebuked for pointing out this simple truth, but judging from the wealth of the literature we have available, the hereditarian hypothesis remains to be proven beyond any shadow of a doubt. A series of clever admixture studies could help us parcel out the inextricable social and biological dimensions of race, but to the best of my knowledge, none exist up to the time of this writing. We at least owe it to future generations to find out exactly what is responsible for gaps in academic performance, both within populations, and between populations (don’t say it’s all poverty and discrimination, because we know that’s not so), and how they may be mediated through environmental interventions that are well within our grasp. If that means billions of dollars in additional welfare spending, so be it. The opportunity costs of not doing so are simply too high.

    And if it turns out that some people, through no fault of their own, cannot be rescued under any circumstances*, then we ought to provide them a niche in our society where they may live out their lives in dignity. “High IQ genes” or not, there is no reason why anybody should suffer a poor standard of living due to circumstances that are largely beyond their control. To insist otherwise would be immoral.

    Lastly, as for the idea that some people score low on academic tests because they just don’t care, even when it is well established that scores on these tests can have lifelong consequences — yeah, sure, buddy. I suppose that is exactly the same as owning a ferrari that is invisible to everybody but yourself. It might as well not even exist. What’s the damn point, really?

    _____________

    *We know such people exist, already! Being wealthy, white, or heterosexual won’t help your career if you suffer from familial mental retardation. And yes, we have identified many of the precise genetic defects that are responsible.

    IQ, as you mentioned, exists on a spectrum, with all sorts of gradations in between “genius” and “mentally retarded”.

  7. God, that post was a lot longer than I expected it to be. Don’t bother reading it if it is too cumbersome.

  8. I missed a couple of points, too, but that will have to wait until later. That is, if anybody even bothers to read my verbiage.

  9. “Suffice to say there is a long body of work within anthropology (and cross-cultural psychology)…”

    LOL… at that point I had to stop taking you seriously.

  10. @misdreavus Thanks for the thoughtful reply-there are parts I can agree with, and parts not (BTW I read down the whole way!). I’ll get back to you soon.

    @dvz. I don’t understand your point. It would be better perhaps if you addressed what Michael wrote. You might start with this:

    “…I will raise three lines of objection to Hsu’s claim to have found the “genetic architecture” of high IQ. The first objection is that the tests Hsu uses to determine IQ are culture bound. That is, they reward some types of test takers more than others. Second, Hsu’s model assumes that certain types of knowledge require more cognitive ability to master than other forms do. Third, Hsu falls victim to the utilitarian fallacy. I’ll briefly discuss the first two before turning to the third….”

    @Razib Khan: I don’t know why you find Michael’s blog a “Bizarre attack.” I thought it was pretty reasoned and timely, particularly in the context of what has come out of Jason Richwine’s PhD. dissertation as well as his work at he Heritage Foundation.

  11. @Razib. I guess we have different types of humor! Funny never really occurred to me–preachy perhaps? Tendentious maybe? But funny? I don’t think Jay Leno has much to fear.

  12. preachy + not-even-wrong = funny

    (at least in cases where people aren’t getting hurt; e.g. anti-vaccination crap)

  13. @misdreavus

    You write:

    “it becomes an inescapable conclusion that as crude as IQ tests are, they measure some property (or linear combination of properties) of the nervous system that corresponds to what we call “intelligence” in the context of modern civilization”

    The problem with your assertion, which you also make up front, is that modern civilization is a context too broad to be meaningful. Once you start unpacking “modern civilization” you will find the predictive ability of IQ falls apart.

    @razib

    People do get hurt in arguments over IQ. And just as with anti-vaccine crap, those on the receiving end of quackery suffer. And make no mistake, the “g-factor” is the herbal medicine of social science.

  14. ***This is a well understood phenomena. The more money one’s parents make (or the more academic their jobs are!), the better one’s chances of succeeding on the SAT.***

    Yes, but if cognitive ability is significantly heritable (molecular genetic studies, plus twin/adoption studies show it is) then you would expect the children of academic parents to have similar abilities. Just as you would expect children of tall people to be taller than average.

  15. ***And make no mistake, the “g-factor” is the herbal medicine of social science.***

    This undermines your credibility somewhat. The data shows it works pretty well (Steven Pinker ‘My Genome’ NY Times 2009). In fact, a more reasonable claim would be it is the most useful tool in the social sciences.

  16. @ J Steinberg

    Or, that IQ tests measure social position as indicated by the ability to excel at standardized testing.

    In the mid 1950’s standardized testing in the US was about a 7 million dollar industry. Today it is worth 500 million or so. On top of this you can add prep companies, etc. It is now the tool of social mobility in the US.

  17. ***that IQ tests measure social position as indicated by the ability to excel at standardized testing. ***

    Note though that identical twins raised apart tend to have very similar scores, while there is virtually no correlation between adoptees and their adoptive parents or siblings.

    Also, notes molecular genetics studies like this one which are confirming the findings of various twin and adoption studies.

    http://www.nature.com/mp/journal/v16/n10/full/mp201185a.html

  18. IQ tests come a lot closer to measuring how good you are at algebra than your social position – IQ is only moderately correlated with SES, more’s the pity.

    Consider the chance of someone doing well in a math course, Someone with an IQ of 130 with blue-collar parents (which does happen) is far more likely to do well than the child of a Fortune 50 executive with an IQ of 110. There is no royal road to geometry.

  19. Some nice examples of deep play with the rules of standardized testing can be found here: http://www.testingmom.com/

    Take a look at the testimonials. For example:

    “LOVED the experience we had with your site! Our boy scored in the 99th percentile on both the OLSAT and Bracken test. Thank you TestingMom.com!!!”~ Xiong and Lin- Dad and Mom, Queens, New York

    Who games the test through metacoginition? Mom. She will get you into preschool. Think of her as the anti-Feynman.

    Happy Mother’s Day!

  20. The problem with that study is the age of the twins is 7. A consistent finding has been that heritability rises with age. It would be interesting to see a follow up on the twins in adulthood.

    There’s a more recent one here which discusses the environmental impact of low SES. They conclude:

    “The notion that heritability may be lower in lower-SES families is appealing, in part because of its environmental implications: If heritability is lower in lower-SES families, it suggests that environmental interventions might be more effective in boosting cognitive development for children in lower-SES families. The present study, which is based on a large UK-representative sample of children followed longitudinally, leads to a similar implication. Although the genetic influence on IQ is the same in lower-SES families, shared environmental influence appears to be greater in lower-SES families, suggesting that family-based environmental interventions might be more effective in these families. However, two further aspects of the results temper the policy implications of this finding. First, shared environmental influence is found in both lower- and higher-SES families and the difference in shared environmental influence between them is modest. Second, shared environmental influences on IQ decline from childhood to adulthood so that these influences might not have an impact in the long run.”

    http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0030320

  21. @ gcochran

    You overlook the case of “being good at algebra” relating to your (potential) social position.

    It is quite possible to improve your test scores (even your IQ score) with some attention and focus. See above.

  22. ***Some nice examples of deep play with the rules of standardized testing can be found here***

    I’m not sure testimonials on a site promoting testing are the most helpful evidence :P

    That’s not to say that people don’t improve, it’s just that the gains appear to be more modest than some might think.

    “After controlling for group differences, the
    average coaching boost on the math section of the SAT is 14 to 15 points. The boost is
    smaller on the verbal section of the test, just 6 to 8 points. The combined effect of
    coaching on the SAT for the NELS sample is about 20 points. The effect of coaching is
    similar on comparable sections of the ACT.”

    http://nepc.colorado.edu/files/Briggs_Theeffectofadmissionstestpreparation.pdf

  23. @ J Steinberg

    This is an interesting question. Quite a few of the studies and reports that claim you cannot raise your test score are sponsored by the testing companies, who obviously have a financial claim on the testing process.

    Then there is an entire industry (such as the website I linked to) which argues that all the tests can be approached as a game and scores raised. They have a strong counter-claim financially.

    You can find studies showing both small and large increases. Quite a bit depends on how far back you go in the prep process. Clearly, for some prepping starts before preschool!

  24. @Razib. I hope that you find Max Holland’s post about Evolutionary Psych as amusing as Michael’s!

  25. “The more money one’s parents make (or the more academic their jobs are!), the better one’s chances of succeeding on the SAT.”

    As noted by others, intelligence is heritable and being more intelligent makes it easier to acquire more education and a higher-paying job. Thus it makes sense that SAT scores would be correlated with parental socioeconomic status (SES). The hidden assumptions in the quoted passage appear to be that (a) American society is completely unmeritocratic, with a class structure that is orthogonal to the abilities and skills possessed by individuals, and (b) said abilities and skills have zero heritability. No evidence is presented in support of these extreme assumptions.

    Secondly, as noted by Cochran above, SES is a much less strong predictor of IQ than many people realize. For example, look at this correlation matrix: http://s8.postimg.org/oolz8c0mt/tucker_drob.png It’s from a recent paper by Elliott Tucker-Drob. It shows that while the results of cognitive and achievement tests are strongly intercorrelated (aka the g factor), the correlations between parental education and said tests are modest, ranging from 0.07 (processing speed) to 0.265 (general knowledge). Parental SES is not a very good predictor of children’s IQ because SES and IQ are not perfectly correlated in the parental generation, and parental IQ is not perfectly transmitted to offspring (regression toward the mean).

    “Some skills are more important, and hence likely to receive more development in certain contexts as opposed to others. Hsu is simply mistaking an epiphenomena of formal schooling in the early 21st century for an eternal human truth.”

    You have misunderstood this research. Possessing, say, a certain set of advanced mathematical skills is indeed a cultural epiphenomenon, but individual differences in the acquisition of those skills depend, in part, on cross-culturally invariant differences in neurophysiology. Psychometricians attempt to identify those underlying dimensions of ability differences. For example, factor analysis yields factors that capture information about covariance between different tests. If those factors are reliably linked to variables that are external to factor analysis and the tests used, such as real-life educational and career outcomes, genetic differences, and differences in neurophysiology, then we are approaching those eternal human truths. This is the ultimate purpose of psychometric research on intelligence.

    Similarly, Spearman did not think that classics, pitch discrimination, French, or mathematics were some eternal categories. Rather, he used them because they were, as you say, “highly valued within formal schooling” in that particular historical moment, and thus could be used to reveal individual differences in their acquisition. He used school subjects as indicators of a hypothesized latent variable, viz. general intelligence. Those school subjects were epiphenomena, while the general factor was not.

    According to Spearman, all tasks that require cognitive effort and in which there are individual differences, depend on general intelligence to some extent. Some tasks (e.g., knowledge of classics in his study of school boys) require more of it and are thus better indicators. Subsequent studies have strongly confirmed this principle of “indifference of the indicator” in the sense that performance differences on all cognitive tasks are indeed correlated, and batteries consisting of different tests tap into the general ability. By necessity, most cognitive tests will assess the subject’s acquisition of cognitive skills that are valued in a specific society (his or her own), but in practice these kind of “culture-loaded” tests not only have better predictive power for a wide range of outcomes than any other social science variables, but are also closely linked to genetic differences (i.e., they have high heritability) and a host of other biological variables. But if you want culturally neutral tests, those measuring reaction time and inspection time fit the bill best — these kind of elementary mental tasks load onto the same general factor as IQ tests.

    The existence of the general factor has been consistently replicated across different cultures, using cognitive measures that are appropriate for those cultures. Whether the same abilities are measured across different cultural groups is a purely statistical question. For example, there’s no question that the IQ gap between American whites and blacks reflects a genuine difference in the abilities and skills measured by IQ tests.

    “Hsu simply assumes everyone who takes the test tries to get the best score. He uses the brute fact of admissions based on the SAT to model a market for IQ, which sorts test takers according to their potential with the underlying assumption that test takers do their best to gain high scores.”

    SAT scores are good for his purposes because students have a strong incentive to score well. Differences in motivation may add noise to his data, but they don’t endanger the larger project. Conscientiousness, the suite of traits related to motivation, has a meta-analytic correlation of -0.04 with general intelligence (Judge et al. 2007), suggesting that motivational factors are unlikely to contribute much to IQ differences. If you want to make the case that cognitive test results are confounded by motivational factors, you will have to do much more than present anecdotes about Feynman supposedly scoring “only” at the 95th percentile or whatever.

    “This is an interesting question. Quite a few of the studies and reports that claim you cannot raise your test score are sponsored by the testing companies, who obviously have a financial claim on the testing process.

    Then there is an entire industry (such as the website I linked to) which argues that all the tests can be approached as a game and scores raised. They have a strong counter-claim financially.”

    Nope. Test prep companies report large effect sizes, while peer-reviewed research reports small effect sizes.

    However, even if you could substantially raise scores with test prep it would not mean that the underlying abilities are being improved. A test is not the same thing as the abilities it was designed to measure. You could give the test taker the correct answers beforehand and make him memorize them, so that he’ll get a perfect score. Would that mean that his intelligence increased? Of course not. A test result is not the same thing as the latent ability that is being measured. Teaching to the test can make the test lose its construct validity. Test companies guard their test items jealously for a reason. The larger and more diverse the test is, the more difficult it is to game.

    To make this absolutely clear, an IQ test with, say, vocabulary, arithmetic, and matrix reasoning items is NOT meant to assess the test taker’s vocabulary size or his or her skills at arithmetics or matrix reasoning. The purpose of such a test is to measure latent abilities, such as general intelligence, of which the observable variables are just unreliable indicators.

    You may think that you have a killer argument, but in fact you have not even begun to address anything that is relevant to the study of intelligence.

    “‘I have never worked on anything even remotely related to IQ, so don’t really know what to think about the relation between IQ, immigration, etc,’ Borjas told me in an email. ‘In fact, as I know I told Jason early on since I’ve long believed this, I don’t find the IQ academic work all that interesting. Economic outcomes and IQ are only weakly related, and IQ only measures one kind of ability.'”

    Borjas is a shameless liar: https://occidentalascent.wordpress.com/2013/05/11/whats-in-a-name/

  26. You are pretty wrong in almost everything….
    1 You thinks that high income implies higher SAT(as proxy for IQ) scores but that is not the case. After some important basic levels of general well being (as most people on the USA have as compared to other countries) this implication ceases to exist. In fact what you have is the result of assortative mating. Do a massive study in a country where assortative mating is not so dependent on IQ preference and you will get a lower correlation. Probably very difficult as most societies reward higher iq.. but in the underdeveloped world the pressure is a lot less..
    2 There is a natural order of difficulty in the intellectual pursuits of people. Very few can become theoretical physicists or logicians.
    Also people don’t try hard enough if they are not good enough to begin with(correctly, as they are trying to maximize their utility function), so it’s pretty useless to talk about motivation in the general population or priorities of the social context where they live.
    3 Why people who pay to take a test are not going to be motivated enough as to try to maximize their score?
    About Feynman: I think his verbal iq was not more than 130s, I’ve listened to his talks and videos and he doesn’t seem very impressive. His spatial and math iq probably were very, very high. If that is the case this fact isn’t just in line with Spearman’s Law of Diminishing Returns??

  27. @ Miguel

    Wrong about almost everything? Which part do you think I might be right (or less wrong) about?

    Also, I am curious about your claim (some might call it a bald assertion) that there is a “natural order of difficulty” in intellectual pursuits. Which pursuits are “naturally” more difficult?

    One more thing, just where is the “basic level of general well being” above which the SAT reflects genetic disposition rather than SES?

  28. @ Dalliard

    It was picked up by the spam filter. You wrote:

    “Psychometricians attempt to identify those underlying dimensions of ability differences. For example, factor analysis yields factors that capture information about covariance between different tests. If those factors are reliably linked to variables that are external to factor analysis and the tests used, such as real-life educational and career outcomes, genetic differences, and differences in neurophysiology, then we are approaching those eternal human truths. This is the ultimate purpose of psychometric research on intelligence.”

    If “those factors are reliably linked” then we could follow the logical trail to the El Dorado of eternal human truths. But, the factors are not reliably linked as they lack ecological validity at every point you mention, and a dozen more you leave out.

    If windmills were giants, then Don Quixote would be a hero. Except for the nasty question of ecological validity.

  29. “This is a well understood phenomena. The more money one’s parents make (or the more academic their jobs are!), the better one’s chances of succeeding on the SAT.”

    Of course this is true, because 1) smart people make more money (or have more academic jobs) and 2) smart people have smart children.

  30. Scrogg – i like how you say that herbal medicine is for “crazies” and then say the equivalent of that in this argument is believing in IQ. no, you have to own your own stupid hippie position. herbal is paired with anti-science IQ denier types like your self and real medicine is paired with facts and tests and reality. also, you are truly a despicable, disgusting person. you have zero expertise on anything and can only tear down others like Steve who ARE ACTUALLY DOING SOMETHING and truly moving humanity forward. you are a worthless human being. actually, not even worthless, cuz you’re in the negative.

    you: wring, accomplished nothing, and try to tear others down

    Steve: correct, tenured physics prof., pioneering genetics scientist

    great job!

  31. @ dad

    I am sure Steve Hsu is an absolutely wonderful physicist. I have never offered a line of criticism, nor would I, about his work in physics.

    But, his foray into social theory (and here he is joined by Khan) has only reproduced the same tired old arguments about heredity and intelligence which have been put forward and debunked for over a century. Modulating the argument onto the ground of population genetics makes it neither new, nor insightful. Just more fashionable.

    Further, just as arguments such as these were damaging one hundred years ago, so they are damaging today. The fact that Steve Hsu is an excellent physicist does not excuse his lousy social theory, nor does it automatically confer upon him some kind of magical knowledge of everything in the universe:)

    I am curious about one thing here. When you say people are “truly moving humanity forward” what exactly do you mean? Towards what, for instance.

    And, great job to you!

  32. Towards a smarter world. Just like when you select a sperm donor, you’ll be able to select a smarter embryo…so it can grow up to be a writer on ethnography.com. Do you honestly believe anyone living 1,000 years from now or even a hundred years from now would bat an eye at this concept? One question: you have a choice for a sperm donor between a brain surgeon in America, an “expert” in Koran memorization (like Tony mentioned) from Iran and a super “intelligent” hunter from Africa. Obviously, both of you would just as soon take the African or Iranian as you would the doc, no?

  33. “I am curious about one thing here. When you say people are “truly moving humanity forward” what exactly do you mean? Towards what, for instance. ”

    By finding the gene variants that create IQ differences. This sort of work is opening the door to potentially shortening IQ gaps between individuals, classes, occupations, nations, races, etc.

  34. James reminds me of an excellent point: Hsu is *actually* addressing the problem rather than pretending it doesn’t exist. Saying IQ genes don’t exist is no different than anti-vax people claiming the environment is ruin their kids when it’s actually inherited from *them.* By pretending it doesn’t exist it will remain a problem indefinitely. And what if someone invents a way to improve brain connectivity to “cure” a low IQ? I guess that cure could never exist as there’s no problem in the first place.

  35. @ James
    @ dad

    Suspending reality for a second, assume that IQ differences are genetic and can be controlled. I have two questions for you two.

    First, what makes either of you think the the gaps between “individuals, classes, occupations, nations, races, etc.” would be shortened instead of the opposite?

    Second, what is so wrong with a world full of low IQ people? Why does it need a cure at all? In what way would the world be a better place?

  36. I’m not saying the gaps would close I’m saying there’s a better chance they would close compared to doing it through assortative mating like we do now.
    would you rather live in a world full of IQ 85 people or IQ 145 people?

  37. “But, the factors are not reliably linked as they lack ecological validity at every point you mention, and a dozen more you leave out.”

    Ecological validity is primarily a problem in experimental research. Differential psychology is usually observational, taking place in everyday settings, and causal inference is based on other methods, such as longitudinal and behavioral genetic designs. In the case of IQ, you could ask if test scores have ecological validity in the sense of whether they measure abilities that influence important life outcomes. There is little doubt that IQ is valid in this sense: the validity of IQ is the strongest and most universal of all social science variables. For IQ, there’s an embarrassment of riches of large, representative samples that have been followed, interviewed, and observed for many decades or entire lifetimes.

    “But, his [Hsu’s] foray into social theory (and here he is joined by Khan) has only reproduced the same tired old arguments about heredity and intelligence which have been put forward and debunked for over a century.”

    Steve Hsu has a very good grasp of psychometric intelligence research, whereas yours is non-existent as I showed in my previous post. You construct straw men and then attack them with anecdotes and purely verbal arguments. That may be the method of choice in your field, but it does not even begin to cut it in psychometrics and behavioral genetics. The research Hsu’s involved with is at the cutting edge of science. It’s your arguments that are old and tired; they were debunked generations ago.

  38. Ecological validity is not just a problem of experimental technique. It is a (moral) hazard of all human research.

    The question of ecological validity should bring you to ask: When (in which context) is IQ being displayed? And it should draw you towards the categories employed by the participants, not the categories (life outcome?) deployed by the analyst to measure, poke or, as in the case of IQ, close off potential futures.

    Are you at all familiar with the debate between Wundt and Kulpe or the Torres Straight Expedition?

  39. I have no idea what you’re talking about. Please spell out your argument. Present the evidence that supports it with regard to IQ, and also specify the kinds of evidence that would falsify it.

  40. Some subjects really are harder than other subjects. Do you deny this?

    Do you think the millions of genetic variants have no impact on cognitive function even though they have impact on all the other parts of the body?

  41. @ cheivous2013

    I looked a few days back and I didn’t see it.

    @ Randall Parker

    You are asking the wrong question, for a start. Are some subjects harder than others? For whom and under what circumstances?

    There can be no doubt genetic variants have an impact. The question, again, is how much and in which circumstances?

Comments are closed.