Tags
Few traits affect our self-esteem as profoundly as IQ and income, so suggesting that the two are correlated makes a lot of people unhappy. While many people will concede that such a correlation exists, they will insist on making certain caveats, such as that IQ is only important below IQ 120 (see Malcolm Gladwell) or that IQ is only important because it reflects the social class you were raised in. Even the Lion of the Blogosphere, got in on the act, writing:
While noting that IQ is as unimportant as it is, it is probably even less important for people who go to college, because low IQ is a better predictor of poverty than high IQ is a predictor of financial success. In fact, there is other evidence out there to suggest that, after educational credentials are accounted for and for those who have at least a college degree, higher IQ is actually correlated with lower income. (The corollary is that IQ is a useful predictor of income for people who do not have a college degree, because people without college degrees but with higher IQ are more likely to wind up in skilled blue-collar careers. For example, there are many civil service jobs that pay decent middle-class salaries which use g-loaded tests for hiring purposes.)
But is this true? In a fascinating 1996 analysis (see Figure 2) people with a four year college education were divided into five categories based on cognitive ability, with level 5 being the brightest and and level 1 being the dullest, and their weekly wages reported. The result was that college grads with level 5 intellect typically made $631 a week, level 4 made $561 a week, level 3 made $503 a week, level 2 made $466 a week and level 1 made $357 a week. So even among people who all have a college degree, there’s a very clear positive meaningful correlation between IQ and income.
A similar pattern was found among those with just a high school education, except here incomes were depressed for all intellect levels, showing that education affects income independently of IQ. With only a high school diploma, level 5 intellects now typically made only $447 a week, level 4 made $340 a week, level 3 made $320 a week, level 2 made $294 a week, and level 1 made $265 a week.
So while high IQ helps makes you richer, regardless of whether you have a college degree or not, the Lion is also correct to emphasize the importance of college. College grads make more money than high school grads, even when they’re three levels dumber.
jorge videla said:
i think that’s only when the full range of iqs is looked at.
obviously a grad with an 85 iq will do less well. but hardly any are that dumb.
and even though american unis select for iq less than other countries’ unis, elite schools do have much smarter than average grads.
jorge videla said:
level 5 is where almost all grads are. right? so it’s a silly study.
the general societal survey showed a correlation of 0 or slightly < 0, where wordsum was the iq proxy.
pumpkinperson said:
Unfortunately I don’t know what IQ scores the different levels correspond to but I think it’s unlikely that almost all college grads are at level 5, because as you say, that would make the study pretty silly.
As for the GSS research, vocabulary is a good proxy for IQ, having a g loading of at least 0.7, but that’s partly because vocabulary tests are long (typically including several dozen items) giving them a high split-half reliability coefficient. That reliability likely plummets on a vocabulary test that’s only only 10 words long,
Wordsum is probably an especially poor discriminator at the upper end because a perfect score of 10/10 gives an IQ of 128. Get just 2 words wrong and your score drops to, I believe, 113. On the WAIS-III, the average college graduate has an IQ of 113, so no wonder it doesn’t correlate with income among college grads. The difference between an 8/10 and a 10/10 is probably mostly luck (the test just happens to sample a word one is familiar with).
The fact that the average mathematician scores 117 is a red flag.
http://anepigone.blogspot.ca/2011/01/average-iq-by-occupation.html
Jensen reported that the average mathematician has an IQ of 143! I don’t know if they’re that high on average, but I think they’re substantially higher than wordsum suggests.
pumpkinperson said:
I don’t know if American universities select less for IQ than in other countries since America uses an especially g loaded test for college admissions (SAT). And yes, the average IQ at the most elite schools is extremely high (about 2 SD above the U.S. mean):
jorge videla said:
you’re right. american unis are the only ones which us an outright iq test. but this is only one factor, and all the other factors are not very g-loaded.
but other countries use cumulative exams as the only criterion for admission. a-levels in the uk are taken together an iq test and they’re the only criterion.
http://en.wikipedia.org/wiki/Intelligence_quotient#Social_outcomes
but then there’s this: http://www.independent.co.uk/news/iq-tests-should-replace-exams-1588320.html
in fact ’50s britain was much more economically mobile than today’s britain, because smart but poor kids were identified by the 11+ exam and went to state grammar schools. these have been done away with since. somewhat paradoxically, perhaps, when people are selected for iq alone, the result is greater economic mobility.
america’s best unis have a lot of the very smartest people, maybe half. elite unis in other countries have all of them.
but canada is even worse than the us, as they select on “grades” alone.
jorge videla said:
so the liberals’ opposition to tests as sole criterion in america is exclusively motivated by the lower scores of non-whites, non-asians.
but the result is lower economic mobility!
i support affirmative action based on class. and so do many republicans (i hate republicans, btw). the reason why the dems are opposed is that such aa would overwhelming benefit “poor white trash”.
jorge videla said:
The largest area still operating the “Eleven Plus” after the system was phased out in Northern Ireland in 2008[4] is the county of Lincolnshire (although the test is optional, the education system is completely Tripartite- every major town has Grammar and Comprehensive/ Technical Schools).
so if i’m right you’d expect to see lincolnshire’s lower class children over-represented at the highest level.
pumpkinperson said:
The SAT correlates about 0.8 with self-described IQ tests, corrected for range restriction. Grades correlate about 0.65 with self-described IQ tests corrected for range restriction. a-levels in the UK are probably somewhere in between. So you would expect elite U.S. universities to have smarter students than elite Canadian/British universities. But as you say, U.S. universities seem to give disproportionate weight to non-academic criteria (sports, legacy status, etc) so it probably roughly balances out
But it bothers me when someone’s high score on the SAT gets them into Harvard which then allows them to get rich or powerful, because then it was the IQ score that caused the success, and not the IQ itself which caused it. It’s much more interesting when a high IQ person skips college but gets rich anyway because they actually did something intelligent like spotted a business opportunity or invested wisely or invented some technology. Smart people should rise to the top naturally, not because they did well on some college admission test. g should cause success directly, and not because the SAT or even grades served as an intermediary.
Brain size tripled as apes evolved into humans, not because apes had to pass an SAT test to survive, but because life itself was an SAT test. Universities should stop trying to be gate keepers and just let smart people rise to the top naturally.
jorge videla said:
Grades correlate about 0.65 with self-described IQ tests corrected for range restriction.
everything correlates when “corrected”. and which grades? not hs or college grades. not what i’ve seen at all. maybe first grade grades?
i gave you the link…taken together gcses (not a-levels, my bad) have a g-loading the same as the sat, .8. of course the g-loading will depend on the battery and the population, but they’re basically the same.
that is, from a psychometric pov, british and other countries’ admit based on iq only. the president of iran, believe it or not, scored first or third in iran on some entrance exams and his dad was a blacksmith. the rest of the world’s elite is smarter than america’s and canada’s. it’s just math.
jorge videla said:
sorry about that ‘countries'” apostrophe. one anticipates and then changes his “train of words”.
BUT. dear God, if admissions, hiring, and preferment of all kinds were transparent, the result would be 1. much more mobility and 2. much more redistribution.
buffett is prescient on this, just as he has been on investments.
i’m an american, born and bred (as st paul might have said), but the frontier is over. 313 million people…there’s no frontier. yet many americans still think as if there were randian supermen and men who pulled themselves up by their own bootstraps.
the ideas of conservatives are attractive…but they don’t correspond to reality.
my great great granddad was a self-taught mining engineer and husband of the Titanic’s most famous survivor, but those days are OVER.
jorge videla said:
well i showed my bias on that one!
According to his autobiography, he was ranked 132nd out of 400,000 participants that year…
but anyway, as Blues Traveler said, that supposed nutcase Mahmoud Ahmadinejad very likely had a higher iq than any high ranking american politician.
pumpkinperson said:
Yes, the 0.65 correlation between IQ and grades is in elementary school where you have virtually the full range of IQ in one class room. The correlation drops to about 0.55 in high school and about 0.4 in law school, but the further you move up the education ladder, the more range restriction you get. You have to adjust for range restriction because the lowest IQ in a high school calculus class if often 115 which spuriously lowers the correlation because there are not enough IQ differences for grades to correlate with. SATs only correlate well with self-described IQ tests after range restriction adjustments.
Grades would correlate even more with IQ if they did not give so much weight to factors like attendance, completing assignments, etc. If grades were just based on whether you learned the material, they would be even more g loaded. Perhaps in other countries, college entrance exams are indeed based on how much of the material you actually learned in high school, so as you suggest, those exams might be very highly correlated with IQ.
You’re probably right about the elites in other countries being more highly selected for IQ (relative to their population mean and their population size). Another factor in this that a scholar named DK Simonton seems to write about is that in the U.S. the general public votes on the president so a presidents needs to be smart enough to impress average Americans, but in a parliamentary system used in other countries, the prime minister must be smart enough to impress other members of Parliament, which presumably takes more IQ:
http://books.google.ca/books?id=V8Sas74bBjwC&pg=PA236&lpg=PA236&dq=simonton+parliamentary+iq+prime+minister&source=bl&ots=M5dP32V0x6&sig=URQvy5yGkQSGHdoyuFPmiDeGBjY&hl=en&sa=X&ei=YeTRU4OVMcefyASukoKYDg&ved=0CB4Q6AEwAA#v=onepage&q=simonton%20parliamentary%20iq%20prime%20minister&f=false
pumpkinperson said:
my great great granddad was a self-taught mining engineer and husband of the Titanic’s most famous survivor, but those days are OVER.
Are you saying that your great great grandmother WAS the Titanic’s most famous survivor?
jorge videla said:
there’s a correlation, but it’s weak and whatever it is it would hardly be a legitimation of america’s enormous inequality and inequality of opportunity.
in denmark and the netherlands father’s income is basically not correlated with son’s. in america father’s income is a better predictor of son’s income than the son’s iq.
denmark and holland are nice places.
america is a satan’s own a-hole.
pumpkinperson said:
The correlation between IQ and income is about 0.4. Controlling for education probably reduces it to 0.2 which is weak, but even weak correlations can have big effects at the extremes. Also, the reduced correlation is largely just a function of range restriction. Among the cognitively homogeneous, it’s hard for IQ to correlate well with anything since the range of IQ has been truncated.
Whether America’s enormous income inequality is justified probably has more to do with the distribution of intelligence than the g loading of income.
Most people assume intelligence has a Gaussian distribution, and indeed IQ scores are forced to fit a Gaussian distribution with a mean of 100 and an SD of 15.
As for income, one study found that in a large sample, U.S. wealth had a mean of $145 K and an SD of $448 K. If wealth had a Gaussian distribution,the richest American (Bill Gates) would have a net worth of only $2.6 million. Instead he’s tens of thousands of times richer, so income inequality seems unfair if wealth differences are so much greater than intelligence differences.
But one radical possibility is that intelligence differences are vastly greater than IQ scores imply. Indeed a Promethean once suggested that problem solving speed doubles every 5 IQ points:
The other factor you must keep in mind is that differences in wealth (like any other achievement) are probably multiplicative, not additive. So if your 1 SD above average in IQ, 1 SD above average in work ethic, 1 SD above average in luck, 1 SD above average in looks etc, your expected success might be something like:
1 x 1 x 1 x 1 = 1
But someone who is 4 SD above average in all these dimensions might have the following success:
4 x 4 x 4 x 4 = 256
So that might also partly explain why some people are many many times richer than others.
A multiplicative model would also suggest that no matter how high your IQ, if you’re a zero on any other important trait, your success will be zero, since anything multiplied by a zero is a zero. That would explain why some incredibly brilliant people are dirt poor.
I’m not saying this is how it should be (politically I’m quite indifferent). Just trying to make sense of things.
jorge videla said:
yes. obviously very rich people tend to be very smart. even when all they’ve done is inherit.
my point was…
even if there were a perfect correlation, inequality at american levels is still unjustifiable.
conservatives never take that next logical step.
warren buffett has…
i can’t find the quote, but buffett says those with high ability should regard themselves as lucky…just as if they’d been born rich.
i guess i’d say i’m an anti-authoritarian lefty, but i believe in eugenics. in fact, i think the deep reason why eugenics is “unpopular” is that it would immediately invalidate the “you know those people…” argument for gross inequality. neo-liberal policies couldn’t survive serious eugenics.
pumpkinperson said:
i can’t find the quote, but buffett says those with high ability should regard themselves as lucky…just as if they’d been born rich.
Then maybe we should tax people based on IQ instead of on income. It would be pretty funny to see some liberal upper middle class professor with a 140 IQ forced to pay more taxes than some IQ 90 pizza entrepreneur gazillionaire.
I believe a major reason why so many academics are liberal is because they resent the fact that many people less educated or less intelligent than themselves are so much richer. A lot of liberalism is rooted in elitism and snobbery. They’re mad that the wrong people are getting rich.
jorge videla said:
A lot of liberalism is rooted in elitism and snobbery. They’re mad that the wrong people are getting rich.
that’s absolutely true. but they’re half right. making a gazillion in pizza you’ve created nothing. and at the same time most very high iq academics will never do anything worth gazillions.
us and them, either/or, is false.
the great progress that’s been made since 1914 or 1918 is due entirely to technology. whatever “system” most rewards engineers and scientists will be the richest. neither capitalism nor communism does this.
the soviet union was a failure? errr…brainwashed person! the russian empire was a backwater in 1917. it was like brazil. its “failure” is better explained by hbd and environment than economics. but north korea is fubar.
jorge videla said:
and you really should read Xenophon’s Oeconomicus. http://en.wikipedia.org/wiki/Oeconomicus
capitalism is nothing new. what’s new in late 18th c England is the steam engine and the power loom. technology trumps economics.
thankfully i live in pdx, so i can check it out…public library systems elsewhere in the
us suck john holmes.
jorge videla said:
http://www.businessinsider.com/warren-buffett-on-the-ovarian-lottery-2013-12
pumpkinperson said:
Yes, the pizza gazillionaire probably does not contribute an outstanding amount to society, but ironically, the academics who most resent his success tend to be the ones in the most useless academic fields, so coming from them, the anti-capitalist tirade is hard to listen to. The academics is useful fields like STEM tend to be (I assume) more free market libertarian, which is ironic, because as you say, technology types contribute the most to society, so they should be the most resentful about not getting a bigger slice of the pie, but their love of mathematical systems (capitalism) and metaphorical Darwinism overwhelms their personal greed and ego. And I know you also argue that libertarians tend to have aspergers, and perhaps that’s true too in the sense that they are naive to the abuses of capitalism, since I’ve previously argued there’s been an evolutionary trade-off between social IQ and technological IQ.
I do agree that a major flaw in capitalism is that engineers and other high tech innovators are not rewarded commensurate with their contributions. For example, in the past, a gifted author could only write one copy of her book by hand, and thus was extremely limited in how many copies she could sell.. But because of technology types inventing the printing press, the word processor, and the photo copy machine, authors can sell millions of copies and become gazillionaires. But really, it’s the inventors of those technologies that made it possible for so many people to read the book, yet they often get very little of the profit. And similarly, singers and movie stars get wildly rich because technology they didn’t invent allows them to reach millions of people they could never have reach in past centuries.
jorge videla said:
i forgot to mention steve hsu’s data on the u of oregon.
the u of o is a miditier state u. its sat distribution is the same as the sat distribution as a whole. it does have some very high scorers.
what prof hsu found was a correlation of .4-.5 both overall and within major between gpa and sat, both composite and math. you can look it up on his blog. he’s got graphs. iirc, there were a lot of 1500+ with gpas < 2.0.
correcting for restriction of range in this case is inappropriate. but, yes, if corky attended he’d likely have a really low gpa.
pumpkinperson said:
In my humble opinion, one might still want to correct for range restriction, because while the u of oregon might be representative of all college bound American students across the full range of universities, it’s likely not representative of all American 18-22 year olds, since many of the dullest are not academically ambitious enough to even take the SAT in the first place. Just the act of taking the SAT implies an above average IQ.
One would have to convert the SAT scores of u of oregon students into IQ equivalents and then calculate the standard deviation of the students. If it what was smaller than 15, then range restriction adjustments would be considered appropriate.
There have been studies where the opposite occurred. I seem to recall one study (don’t have a citation) that found an unusually high correlation between IQ and MRI brain size, but the sample consisted of mostly mentally disabled and gifted people with almost no one in between. Thus the authors corrected the correlation for extended range, thus reducing the brain size IQ correlation from something like 0.6 to something like the more typical correlation of 0.4.
jorge videla said:
correction for restriction of range regarding college grades makes no sense unless the “correct” range is the range of those who attend college rather than the range of the population as a whole. it’s the same reason that the supposed correlation between income an iq for grads is meaningless when the range is the full population range, most grads are in the top quintile.
g is what’s heritable in iq, yet the higher the iq the less it’s explained by g. this is spearman’s law. even the linear model phenotype = G + E predicts that very high iqs are more explained by environment than middling iqs. very high iqs are mostly NOT the result of genes.
lotb’s post on “career tracks and merit” does describe how the real world works. i know, because i’ve witnessed it first hand. in fact, it’s even worse than he describes, because there is also discrimination based on class background. if you don’t sound like the right sort, “fuhgetaboutit”. this discrimination may not be deliberate, but the children of the upper middle class sound different even when the’re not as bright as the rest.
the idea of a world where raw ability determines accomplishment is ridiculous. man is a social animal. he is the social animal. and production is more social than ever. a few hundred years ago most were farmers. the unit of production was the family. few ever met a stranger, let alone were forced to work with people for years whom they would never know well.
over the last 250 years in western europe and its diaspora and later in japan and s korea etc. production has been socialized, yet many still persist in their adherence to the individualist ideology.
jorge videla said:
correction for restriction of range regarding college grades makes no sense unless the “correct” range is the range of those who attend college rather than the range of the population as a whole. it’s the same reason that the supposed correlation between income and iq for grads is meaningless when the range is the full population range, most grads are in the top quintile.
g is what’s heritable in iq, yet the higher the iq the less it’s explained by g. this is spearman’s law. even the linear model phenotype = G + E predicts that very high iqs are more explained by environment than middling iqs. very high iqs are mostly NOT the result of genes.
lotb’s post on “career tracks and merit” does describe how the real world works. i know, because i’ve witnessed it first hand. in fact, it’s even worse than he describes, because there is also discrimination based on class background. if you don’t sound like the right sort, “fuhgetaboutit”. this discrimination may not be deliberate, but the children of the upper middle class sound different even when the’re not as bright as the rest.
the idea of a world where raw ability determines accomplishment is ridiculous. man is a social animal. he is the social animal. and production is more social than ever. a few hundred years ago most were farmers. the unit of production was the family. few ever met a stranger, let alone were forced to work with people for years whom they would never know well.
over the last 250 years in western europe and its diaspora and later in japan and s korea etc. production has been socialized, yet many still persist in their adherence to the individualist ideology.
pumpkinperson said:
Correcting for range restriction is always an arbitrary decision, of course. If you want a realistic estimate of how much IQ influences outcomes within some specified population (i.e college grads) then you should not correct for range restriction, but it’s useful to correct for range restriction to determine whether a g loading is low because the criterion is intrinsically unrelated to IQ, or if if it’s just spuriously deflated by too few IQ differences for the criterion to correlate with within the sample studied. For example, you might find that there is a higher correlation between IQ and learning to ski then there is between IQ and learning theoretical physics, but it would be absurd to conclude that theoretical physics is less cognitive than skiing because the latter correlation was spuriously depressed by the fact that the range of IQ among physics scholars is extremely restricted while the IQ range of skiers is wide, so correcting for range restriction allows for apples to apples comparisons when such comparisons are desired..
I’m impressed by your level of psychometric knowledge. You’re absolutely right that the more extreme the IQ (high or low), the greater the discrepancy between IQ and genotypic IQ. Of course that’s just regression to the mean and would happen for every heritable trait. The more extreme your height (tall or short), the greater the discrepancy between your height and your genetic height. So this doesn’t imply any unique flaw in IQ measures. IQ never claimed to measure only the genetic component of intelligence, though it has historically claimed to measure innate capacity, though there’s a difference. Fetal Alcohol Syndrome is innate but entirely environmental.
Spearman’s Law of Diminishing Returns is an interesting phenomenon though I’m a little skeptical of its authenticity because if g were really so diminished at high levels, why do so many people who score 700+ on the verbal SAT, also score 700+ on the math SAT? If it is authentic, then yes, that would seem to imply that extremely high IQ’s have very little genetic component especially when combined with the regression effect. It would be interesting to look at identical twins reared apart to see if the twins of extremely gifted people regress more to the mean than predicted from standard genetic formulas based on the population as a whole.
The Lion’s career track theory is very good and I think he does a great service to young people by warning them to get the right credentials because our society does give prestigious degrees disproportionate weight. I just think he overstates his case by implying there’s NO correlation between IQ and income within career tracks/education levels, or he sometimes claims there’s even a NEGATIVE correlation. The problem is that the correlation between IQ and real income is so non-linear that it’s very hard for people to observe among their coworkers, friends and family. I estimate that for every ten fold increase in financial success, average IQ increases 8 points.
And among people with identical education/career tracks, it’s probably closer 4 IQ points per ten fold increase. For most college grads, everyone they know makes either five figures or six figures, so this pattern would be hard to subjectively observe.. However if you were to meet a bunch of college graduate homeless and college graduate billionaires, the IQ difference between them would be unmistakable.
Merrill Babilon said:
I was reading through some of your content on this internet site and I conceive this web site is real instructive! Continue posting .