IQ and achievement: pinker is kinda right and kinda wrong
In the New York Times, there is an essay by Steven Pinker, who goes after Malcolm Gladwell. Fair enough. Gladwell’s a great writer, but he’s a journalist who loves telling a great story, so he makes an easy target. But I was a little confused after reading this passage:
It is simply not true that [stuff that Pinker accused Gladwell of getting wrong, but Gladwell has a strong defense*] or (the major claim in “Outliers”) that above a minimum I.Q. of 120, higher intelligence does not bring greater intellectual achievements.
I was a little struck by this. According to Keith Sawyer’s textbook on creativity, Pinker is wrong. Sadly, I just returned my copy to the library, but my memory is that high achievement in creative areas is not always tied to IQ.
A quick google scholar does back up my memory. For example, the Journal of Creative Behavior has a recent article on a meta-analysis of the correlation between divergent thinking (the ability to conceptualize novel ideas), IQ and achievement. The paper shows that the literature is far from settled on the issue, and that a meta-analysis finds a modest IQ/creative achievemnt correlation, but the correlation depends a lot on specific instruments used. The paper also claims that divergent thinking has a modestly larger correlation with achievement than IQ.
Now, I think that Pinker is still correct in a very important sense. If you look at the entire human population, IQ is going to be a huge predictor of success. Nearly every regression of income and educational attainment that has IQ as a regressor shows a positive correlation. Though I haven’t seen any studies, I’d be shocked if IQ, in the whole population, didn’t predict artistic achievement. One needs substantial intellectual capacity to make a film or write a book. There’s not going to be a threshold effect either.
But I think Pinker may be missing Gladwell and Sawyer’s point: within professions, IQ is probably a weaker predictor and maybe has no link. Why? It’s because high achievement is a multi-stage process. Yes, you need IQ to become a mathematician, but success in scientific and artistic fields also depends on attributes such as emotional control, ability to generate novel ideas, networks, coaching, and academic street smarts. Or, as I’ve said before, you need more than talent. IQ may put you in a position to make an impact, but among people who are in that position, success may be determined by things other than raw analytical capacity.
* Originally, I took Pinker’s word on stuff like the quarterback issue, but if you read Gladwell’s response, the man’s got a point.
orgtheory.net 
Gladwell has a good response to Pinker (though not on the IQ issue in particular, but still about achievement) posted at his own blog today: http://gladwell.typepad.com/gladwellcom/2009/11/pinker-on-what-the-dog-saw.html
I think he’s quite a bit more generous to Pinker’s questionably-sourced criticism than he needs to be.
LikeLike
Alan
November 17, 2009 at 1:38 am
I guess my question is, does Gladwell (or Sawyer) make the strong claim that above 120, IQ is not a contributor to achievement, or the far more modest claim that other factors are more important?
LikeLike
Michael Bishop
November 17, 2009 at 1:43 am
I was also a little surprised at Pinker’s claim that offsetting behaviour (risk compensation) was ‘demonstrably false’. My impression was that we don’t necessarily increase our risk taking in perfect one-to-one measure with the diminishment of risk in other areas (i.e. risk homeostasis), but that there was some decent evidence suggesting that we do a little bit of risk compensation, taking new risks when old ones are controlled.
I don’t see the basis for Pinker’s claim that offsetting has been debunked: Jackson & Blackman (JAP, 1994) and Stetzer & Hofmann (OBHDP, 1996) both seem to demonstrate that we do compensate for reduced risk by taking increased risks in other areas (though not to the point of entirely swinging back to the original level of risk).
LikeLike
Lukas Neville
November 17, 2009 at 1:58 am
The importance people place on IQ seems so strongly correlated to political viewpoints. And not in the usual way–if you believe the right that IQ is everything; then we don’t really have a very meritocratic or equality of opportunity society, and should probably do a lot more redistribution to help the cognitively inferior.
LikeLike
Thorfinn
November 17, 2009 at 2:45 am
It’s worth remembering that the IQ test was developed to determine which children needed extra help to perform at grade level. It was not developed to predict anything about high performers or to make claims about ultimate ability.
LikeLike
Mikaila
November 17, 2009 at 5:14 am
[…] of IQ, specifically, and intelligence, broadly, in success. Recent research has generally shown little link between intelligence and success within fields, and that there are multiple kinds of intelligences that drive achievement. On the other hand, […]
LikeLike
Two men with big hair and the nature of IQ « Dance4
November 17, 2009 at 5:55 am
The expertise literature is highly relevant to this debate (http://bit.ly/Y8or) as is the selection and recruitment literature.
The expertise literature typically shows that years of deliberate practice is the main differentiator in domains like chess and violin playing (for some interesting empirical examples, see Ericsson et al’s 1993 Psych Review article available here: http://bit.ly/1Xyk).
The selection and recruitment literature suggests a strong relationship between intelligence and performance. Schmidt and Hunter’s (2004) large-scale meta analysis found correlations around r = 0.4 to 0.6 between intelligence and job performance. They also report large differences in average IQ between professions (e.g., means: Accountant = 128; teacher = 122; tractor driver = 99.5, etc.).
One way to reconcile the findings of the expertise and the selection and recruitment literature is to acknowledge that expert performance is largely a domain-specific adaptation. Thus, both practice and intelligence are important, but in many domains of expertise variation in practice is so much greater than variation in intelligence.
To take a simple example, a person with a 150 IQ who has played chess for 100 hours is likely to lose to a person with a 90 IQ who has practised in a focused way for 10,000 hours.
Range restriction (http://bit.ly/xHucm) would also be expected to reduce the correlation between intelligence and performance, as would occur if you only sampled people with 120+ IQs. However, suggesting that the correlation would be reduced by range restriction is very different to saying that the relationship is nonlinear (i.e., a positive relationship up to 120 and no relationship afterwards). I find this nonlinear proposition highly unlikely.
A final point, intelligence tests designed to measure intelligence in the general population are typically less reliable in the tails. This would also tend to attenuate observed correlations at the high-end of intelligence, although such attenuation would not be reflective of the true latent relationship.
LikeLike
Jeromy Anglim
November 17, 2009 at 8:57 am
Thorfinn,
What if people credit lucky circumstances, or growing up with good neighbors as the cause of inequality… those would seem to justify more redistribution than if IQ were the main cause.
LikeLike
Michael Bishop
November 17, 2009 at 4:38 pm
Michael,
That’s possible. It seems though that most people observe their partisan heroes, adopt their position on IQ (Gladwell if you’re left; Pinker if you’re right), and rationalize that post hoc. However, I don’t know why Pinker or Gladwell take the stances they do. I’ve met few right-leaning people, even among those who specialize in the field, who don’t believe in IQ, and vice-versa for the left.
LikeLike
Thorfinn
November 17, 2009 at 4:58 pm
Lukas,
Could you give us links and/or better explain the studies you mention on risk-taking?
LikeLike
Michael Bishop
November 17, 2009 at 5:09 pm
I don’t find Gladwell’s defense of his quarterback argument to be very strong at all, as the use of per-play statistics as a measure of relative performance is extremely questionable. Mediocre quarterbacks drafted early, payed large sums, and playing for very bad teams are likely to see a substantial number of snaps (think Akili Smith). Mediocre quarterbacks drafted later, in the range that Gladwell claims performs just as well, are likely to see far fewer snaps and in more statistically favorable situations (think Brock Huard).
Using measures like career yards and career pro-bowl appearances, while not perfect, make a good deal more sense, and these measures go against Gladwell’s thesis.
LikeLike
Trevor
November 18, 2009 at 12:33 am
Fabio,
I don’t have the article at hand, but I remember some work by Bowles and Gintis in the past few years that placed IQ fairly far down on the list of the best predictors of income. If I remember correctly, things like parent’s wealth, etc. were more highly correlated with one’s future income.
Has there been better or more recent work that’s taken on Bowles and Gintis’s conclusion?
J
LikeLike
J
November 18, 2009 at 2:16 am
Steve Hsu responds on Gladwell & IQ over 120 here.
LikeLike
TGGP
November 18, 2009 at 7:47 pm
@Michael,
The risk issue (unrelated to the IQ issue, to prevent any confusion) is this:
Gladwell, in his New Yorker essay ‘Blowup’ (http://www.gladwell.com/1996/1996_01_22_a_blowup.htm) reports on risk homeostasis theory:
“Under certain circumstances, changes that appear to make a system or an organization safer in fact don’t. Why? Because human beings have a seemingly fundamental tendency to compensate for lower risks in one area by taking greater risks in another.”
Pinker responds in his review,
“But as a more substantive claim … that people have a ‘fundamental tendency to compensate for lower risks in one area by taking greater risks in another,’ it is demonstrably false.”
As far as I can tell, the more extreme claims of risk homeostasis theory have not been supported (i.e. perfect compensation), but there is good evidence (the two articles I mention being two examples) that supports some degree of behavioural ‘compensation’ in one area for reduced risk in another.
LikeLike
Lukas Neville
November 18, 2009 at 7:57 pm
@Lukas,
Here is how Pinker quotes Gladwell on risk-compensation: “…that people have a “fundamental tendency to compensate for lower risks in one area by taking greater risks in another,” it is demonstrably false.”
I was unable to access Jackson & Blackman, but I don’t think Hoffman addresses this one way or the other. Because Hoffman tries to assess whether reducing a risk from the environment in one area leads people to be willing to take more risks in the same area. I wish Pinker would offer his evidence that Gladwell is wrong, but I don’t think this counts as evidence that he is right.
LikeLike
Michael Bishop
November 18, 2009 at 10:28 pm
@Michael – Good point. You’re right that most of these studies involve risks within the same broad domain – airbags and speeding being the typical kind of example. But I’m not sure Gladwell’s main argument is about risk crossing domains (he’s talking about safety gains being offset by new risk-taking in NASA engineering projects). And I’m not sure that’s what Pinker’s gripe is about, either.
LikeLike
Lukas Neville
November 20, 2009 at 4:51 pm
“Steven Pinker replies:
What Malcolm Gladwell calls a “lonely ice floe” is what psychologists call “the mainstream.” In a 1997 editorial in the journal Intelligence, 52 signatories wrote, “I.Q. is strongly related, probably more so than any other single measurable human trait, to many important educational, occupational, economic and social outcomes.” Similar conclusions were affirmed in a unanimous blue-ribbon report by the American Psychological Association, and in recent studies (some focusing on outliers) by Dean Simonton, David Lubinski and others.
Gladwell is right, of course, to privilege peer-reviewed articles over blogs. But sports is a topic in which any academic must answer to an army of statistics-savvy amateurs, and in this instance, I judged, the bloggers were correct. They noted, among other things, that Berri and Simmons weakened their “weak correlation” (Gladwell described it in the New Yorker essay reprinted in “What the Dog Saw” as “no connection”) by omitting the lower-drafted quarterbacks who, unsurprisingly, turned out not to merit many plays. In any case, the relevance to teacher selection (the focus of the essay) remains tenuous.”
http://isteve.blogspot.com/2009/11/ny-times-publishes-gladwells-letter-and.html
LikeLike
Josh
November 24, 2009 at 3:30 am