More IQ followup

It has been brought to my attention that Razib from the blog Gene Expression did an analysis of religiosity and IQ by country and discovered that countries with higher average IQs were less religious (the U.S. being an outlier because we are too religious for our IQ).

I discovered the same negative correlation between IQ and religiosity myself in the GSS data. The country correlation lends evidence to the theory that the IQ score differences between countries reflect at least partially the genetic IQ of the countries' inahabitants. And of course it further supports the theory that more intelligent people tend to be less religious.

There is also a post at Gene Expression supporting my discovery that IQ is not related to income after college degrees are factored into the analysis. Michaelv writes, "Further data indicating higher average wages for college dropouts with level 3 NALS literacy than for high school graduates with level 4 NALS literacy strongly suggests that credentialism dominated the determination of socio-economic status."

Are there any studies out there that demonstrate that g has a significant predictive effect with respect to income after college degrees are considered? It seems to me that g primarily predicts income because it predicts the quantity and quality of your educational credentials. This is probably something the IQ community doesn't want to talk about because they are so concerned with proving that IQ matters that they don't want to investigate areas where it should matter but doesn't.

Comments

[omnivore]

As I understand it, IQ does indeed have an effect on income independent of education. The best data I've seen come from the NLSY, and are covered by Murray 1998 ("Income Inequality and IQ", available here: http://www.gnxp.com/MT2/archives/002855.html)

More importantly, credentialism doesn't follow from g's effect on earnings being mediated by education. If society were a flawless IQ filtering system where educational attainment was based solely on cognitive ability, and income were based solely on IQ, you'd observe near-perfect "credentialism" (all income differences mediated by educational differences) despite the utter absence of actual credentialism. In other words, if the IQ test I administer to potential hires causes me to hire smart kids and pay them more, and smart kids also tend to be more educated, I could be accused of basing pay on educational attainment alone (by someone who didn't know my HR practices) -- falsely, though.

[Half Sigma]

omnivore, perhaps I'm missing something in the link you included (which is a fine paper and worth reading).

Murray says that IQ predicts income, and that it predicts income independent of parental SES. The GSS data supports both of those conclusions.

However Murray does NOT say that IQ predicts income independent of education, or if he did I didn't see it anywhere. Murray does specifically say that IQ predicts education and that education predicts income.

And that is exactly what the GSS data showed, that IQ primarily predicts income because it predicts educational accomplishment. But independently of one's degree IQ doesn't amount have much impact on income.

[Half Sigma]

OK, I read the Murray article more carefuly. At page 18 he writes: "The breakdown continues to show a role for IQ in determining occupational status even within groups of siblings that did and dit not get B.A.s. The magnitude of the differences is reduced, however, remaining most conspicuous in the discprency between the Normals and the Very Dull siblings.

Here is my comment:

(1) In comparing IQ (measured by Wordsum) to income, I also found that there was a stronger negative effect for being at the bottom than there was a positive effect for being above average.

(2) My analysis looks at income while Murray looks at occupational prestige. Occupations with higher IQ individuals have higher prestige even though it might not pay any more. For example, a garbage collector might make more money than a college professor even though the college professor job has a lot more prestige.

So people with different IQs but making the same income are probably sorted into different professions, but they still make the same income.

Do you know where I could find the Gottfredson article referenced?

It's frustrating to me that this type of data always counts all schools as equal. The difference between East Coast Private and Podunk State is enormous -- and for all we know, they're counting non-credentialed and online schools as well. As a Podunk State graduate, I'd be hugely surprised if the average IQ there were anywhere near the low-120s.

The author is concerned with medical and engineering schools, but it appears that the data includes all graduate students. That includes child development; family counseling; public administration; heck, probably things like physical therapy and nutrition as well, from state and online schools. I would bet that the latter comprise a much larger share of grad students than do the doctors and engineers.

[Jason Malloy]

Your analysis is interesting and I'm not going to provide a whole lot of evidence here and now, but the truth is more like the opposite. Meaning that credentials are actually worth very little, on average, independent of IQ, and people tend to earn what their IQs would predict, not what their education would.

We see this, for example, with race. Control for education in the NLSY and blacks and whites still have a big wage gap. Discrimination? Control for IQ and the gap mostly disappears.

Meanwhile the theory that IQ stops bringing returns at a "certain level" is debunked by following groups of extremely high IQ people and comparing them with merely high IQ people, as with the Terman and Stanley cohorts. Extremely high IQ people earn even more. We know from personnel selection lit that IQ actually has the strongest relationship with job performance at higher levels of job complexity. See also The Bell Curve for doctors.

[Half Sigma]

Jason,

the doctor example tends to prove my point that it's the credential that matters. All doctors get the same reimbursement from the insurance company no matter how good or how bad they are.

And I acknowledged that my data only goes up to a top bucket of $110,000+ and doesn't tell us how high IQ affects people making super-high salaries.

Within the range of $0 to $110,000+, IQ doesn't have much effect, if any, on income after education is accounted for.

The black white differential you mention also tends to affirm my finding that IQ matters more at bottom but not at the top. In the GSS data, 24.8% of blacks but only 9.6% of whites were in the lowest Wordsum(0-3) category which had the most significant effect on on income.

[Michael Blowhard]

Jason will disagree with me on this -- we've had a few go-rounds about it. But you might (or someone might) want to give some thoughts to the arty fields. I've lived in and hung around these fields for almost 30 years, and I see no connection between smarts and art-talent, at least once a person is decently competent. There are supersmart people with zero art-talent, and supertalented people who can barely think in a straight line. The field couldn't function without some of its participants being pretty sharp -- but these people often wind up in elevated-custodial roles: editors, administrators, board members, etc. The "creatives" themselves range from amazingly quick and insightful to spectacularly dumb (in a brainpower sense). I wonder if it's true that the smarter ones tend to succeed monetarily better, just because they're smarter. Might be. On the other hand, a common thing to see is someone whose smarts either actively get in the way of his talent or who's so prone to overcomplicating that that he messes things up for himself.

My own admittedly dim impression is that art-talent is like athletic talent - it's in the body and the emotions (and imagination and intuitive powers), and has nothing to do with smarts. Having some smarts in addition to a goodly helping of art-talent might be the ideal mixture, but it's one that seems fairly rare.

[Peter]

One field in which it's possible to make significant money, albeit with some risk, is in professional-level sales - selling things like investments, commercial real estate, high-tech equipment, and so on. And it's also my impression that the people who go into that field, while far from stupid, are usually not among the very smartest at their educational levels.

[Randall Parker]

Some jobs have credential requirements for entry. A lot of government jobs are that way. Yes, for them you can expect a strong correlation between credentials and income. But in other jobs credentials matter little.

But how g-loaded is WORDSUM?

[siriata]

Oh my, Halfsigma - Your entire argument has a HUGE fundamental flaw. IQ is positively related to income. IQ is also positively related to academic achievement.... Please let me explain why we cannot conclude that IQ is not related to higher income - or that IQ does not predict income.....

IQ tests today are designed to measure different types of cognitive ability. We have many different cognitive abilities, IQ tests only measure some of them – specifically the abilities that are the core of our education system.. IQ tests focus on language and math skill, spatial ability, pattern recognition, memory and perceptual speed. These are the abilities that are required for the core content in the academic system. Those with greater ability in those areas are more likely to achieve academic success. – As we know, academic success predicts career success/income.

The stats used to test whether a predictor variable (IQ) can predict a criterion variable (academic achievement) is a regression analysis. The same test is used to test whether academic achievement predicts income/career success. This also follows for testing whether IQ predicts income/career success.

If you want to test two predictor variables (IQ and Academic) predicting income/success, the first analysis you would do is a correlation with both the predictor variables (IQ and Academic) – This would show a really strong positive correlation (of course, high scores in either require the same type of skills/abilities)….You cannot conduct a regression analysis with predictors that are highly correlated – because they both measure the same thing!!!!! (duh) – so your fantastic discovery:

“There is also a post at Gene Expression supporting my discovery that IQ is not related to income after college degrees are factored into the analysis.”

Is actually just a lack of knowledge on your part about how to use multiple regression. Thus, if you control for education, of course there is no variability left for IQ – The opposite would also be true. Go try it…..