Sunday, December 21, 2008

Gladwell on Outliers on Charlie Rose

The interview was better than I had expected, but then my expectations were not high. At least Gladwell stops short of completely embracing the "there is no talent, it's all effort" line.

However, at about 13 minutes in we get a huge dose of politically correct pseudoscience and poor logic: Asians are good at math because -- get this -- rice farming was labor intensive. Tolerance for hard work was transmitted culturally and had no impact on genes -- "we know this" says Gladwell :-) I don't suppose Gladwell has looked at any transnational adoption studies, which remove the cultural component...

At 23 and a half minutes we get the "IQ above 120 doesn't matter" claim -- see here for some pretty strong evidence against that.

See here for earlier comments on Outliers, and here for a discussion of success and talent (the meaning of correlation).


Anonymous said...

Gladwell says that "we know" that the difference "asians" (orientals) and western kids (whites) in math is not genetic. The opposite is true. Adopted oriental children have a different intelligence factor profile than whites, namely they score higher in the "viseo-spatial" area, just as Ashkenazim score higher in the verbal factor. They also have bigger brains.

Gladwell is writes what people want to hear and what the prevailing ideology insists on. He is a propagandist.

How can you take a man seriously with an afro who looks and sounds like a woman?

A. said...

"Gladwell is writes what people want to hear and what the prevailing ideology insists on. He is a propagandist."

Gladwell makes a good living this way. Today, after decades of indoctrination, many people are exceedingly uncomfortable with the notion of inherited ability. Particularly if that generalizes to biological differences between the races that have any social significance. Gladwell provides a welcome comfort to those who accept the dogma of multiculturalism. For this he's well rewarded.

Anon-2 said...

How can you take a man seriously with an afro who looks and sounds like a woman?

This is a joke, right?

As for decades of indoctrination, for many more decades in the west, the indoctrination was biased in another direction, mostly at the expense of non-Europeans, and even southern and eastern Europeans. Given that fact, we northern Europeans, who tend to be overly impressed with ourselves, can damn well endure a few decades of re-balancing.

Anonymous said...

The only thing "we northern Europeans" can do better than anyone is play darts.

I wasn't joking.

Don't froget, drug dealers, pornographers, and fast food company executives make a good living giving people what the want. They are no more despicable than Gladwell.

Anonymous said...

See The Flynn Effect: Troubles with Intelligence 2 (via Michael Nielsen).

Steve Hsu said...

The Flynn Effect has been discussed previously on this blog.

Anonymous said...


As far as research goes, I think the only two things important are originality and usefulness. Now at least in my experience (as a PhD student in Computer Science at a big name university), the "smartest" grad students (my defn. would be those that do well in the hardest courses e.g., graduate real analysis, quantum computing etc.) don't end up doing anything terribly more original or useful than those getting borderline A's or B's in the said classes. Once in a while (every 5 years in my guesstimate) some grad student across all CS departments produces truely insightful papers. I doubt every (or even most) PhD in string theory from Princeton (where everyone is probably in the tail) produces work thats original and useful.

If you buy that hypothesis, when relevant coursework doesn't seem to predict success, why should psychometry do any better and why bother wasting time on it?

Steve Hsu said...

CS PhD student:

I agree with you that the difference between an A+ and an A-B in a tough grad course probably does not predict research success very well. Although, if you had a sample of hundreds that you followed for 20 years, there might be a measurable effect. In my own experience the people with enough brainpower to ace the quantum field theory course probably did better on average in their research and future careers than the ones who were a little above average in the course. Certainly, I am more likely to take on a PhD student who did really well in the course, and they know that.

Performance in courses correlates with intelligence and drive, probably less so with creativity. You need all three to succeed -- plus luck! Therefore, we have at least a 3 factor model to consider. Teasing out the effect of one factor may not be easy, but that doesn't mean the factor does not impact success. (Necessary is not sufficient, blah blah...)

Performing reasonably well in a course like quantum computing is already a big restriction of range -- everyone there has been admitted to the big name grad school. In the post above I am referring to the ridiculous statement that IQ over 120 doesn't matter. To get into an *average* grad school in physics or math or CS you probably need an IQ of 130 and to do a theory PhD I would guess you need a significantly higher IQ.

The Roe study indicates that a random sampling of eminent scientists yields an average much higher than 120. (My personal experience agrees with this result.) I don't see a good explanation for that other than that IQ is measuring some quality that is useful for a scientist.

See also Terman's results about stability of adult IQ -- the high IQ kids identified in the study were followed through adulthood. The subset who chose science as a career did not display elevated adult IQs relative to equally bright kids who chose other professions. So, the *causation* seems to be that IQ helps make you a better scientist, not the other way around.

All evidence I have seen suggests that psychometrics actually works -- has predictive power. We use SATs, GREs, course exams, etc. all the time to filter people based on ability and drive. Creativity, as you note, is harder to discern.

Anonymous said...

"Performance in courses correlates with intelligence and drive, probably less so with creativity."

This is an incorrect assumption. The facts are otherwise. Graduate school grades, college grades, high school grades have no significant correlation to IQ. That is, IQ explains at most 10% of the variance in grades beyond elementary school. They would have a strong correlation with cummualtive exams in graduate school and college if there were suhc exams in the US.

They are a measure of pushiness.

In the UK admission to university depends on performance on GCEs and GCSEs which have the same correlation with g as the SAT. The result: in the UK the arrangement of people in the social hiearchy is much more correlated with IQ than it is in th US.

As Noam Chomsky has said, the standard of living in the US is much lower than it should be. Can you imagine if the US had the UK's population density? It would be as poor as Bangladesh. The same goes for Canada which uses no IQ tests (SAT, GRE, LSAT, etc.) to sort people.

Steve Hsu said...

SATs (which are highly correlated with IQ) predict freshman grades in college about as well as high school GPA. That is, a brief exam is as good a predictor of college success as the combined opinions of all of a student's high school teachers.

It's also known that SAT + HS GPA is a better predictor of college success than SAT alone or HS GPA alone. A crude way of describing this is that SAT measures IQ and HS GPA measures a combination of IQ and drive or self-discipline.

These are standard, well known results. Please don't post random assertions unless you have actual data to back them up.

Anonymous said...

I googled "SAT correlation college grades" and clicked on the first thing that came up. Apparently Steve's "standard, well known
results" are NOT.

"Researchers found that the SAT I was by far the weakest predictor, explaining only 4% of the variation in college grades"

I'm sure you take pride in your college grades and the grades you give your students, but their relation to IQ is at best trivial.

"These are standard, well known results. Please don't post random assertions unless you have actual data to back them up."

You haven't given me any "data" either. The article you sight gives no correlation figures. "A strong correlation" may mean .3 or
.5 which means 9% and 25% of freshman grades, and only freashman grades, explained by the SAT.

"The correlation is strongest when all three parts are used, but when each of the three sections is considered singly, writing has the strongest correlation."

And the writing section has the lowest correlation with g. And the new SAT has a lower correlation with g than the old one.

But, of course according to the theme of your IQ posts you may take what I say as gospel. (wink, wink) I took the old GRE once. My score was 800V, 800Q, 800A. Higher than yours I'm sure.

I'm sure you have a VERY high IQ Steve. I'm also dure you are VERY pushy.

Anonymous said...

Give it a rest, guys. It's the holiday season, for God's sake.

Nicolas said...

What is clear is that it is unfortunately another loaded subject, while there is nothing that prevents it from being backed up by data.

Knowing it without a political agenda is crucial for everyone, low and high iq alike.

Not knowing always works to the detriment of everyone, but especially the less favored.

A positive note, is that those are statistics, and individuals are trajectories : everything is always possible !

Anonymous said...

"What is clear is that it is unfortunately another loaded subject, while there is nothing that prevents it from being backed up by data."

For me it is not loaded in the way it is for the typical American.

The problem is with grades, not with the SAT.

In the UK the quality of your college degree depends on two tests. That's it.

Grades in the US can depend on as many as four tests per course per semester, or three per quater plus papers and "home work".

It's rediculous. It's disgusting.
It's America. FAT and STUPID.

Steve Hsu said...

For anyone actually interested in the data, you can find it here and on the College Board site. The University of California more or less replicated these results in recent studies used to decide how to use the SAT in admissions. You can easily find those results as well using Google.

It's important to note that the usual test of the SAT is to predict *freshman* grades. After students self-segregate into different majors (typically with very different SAT ranges within each major), GPA comparisons become less meaningful. Most freshman, however, are taking a common curriculum of courses so the comparison is meaningful. The correlation between HS GPA or SAT and freshman GPA is between .5 and .6. See

Again, a brief SAT exam predicts freshman GPA about as well as the combined opinion of a student's high school teachers (HS GPA). I consider that pretty good given all the complexities.

BTW, has certain political goals that you might want to look into.

G said...

The link to transnational adoption studies you posted seems to me to be commenting on the relationship between different IQ components rather than absolute IQ scores no?

They say "Because of the small sample size and the fact that one cannot rule out some French language handicap owing to early exposure to Korean, one cannot put too much weight on these results." This seems to me to be understating things somewhat - I think this is pretty ambiguous as a result because the adoption age was 18 months on average. One could easily imagine that verbal development was, relative to mathematical, slowed down by this and that overall development still beat the population average by virtue of having 2 college educated parents.

All in all, pretty unconvincing in this instance!

Anyway, Happy Holidays!

Steve Hsu said...

G wrote "pretty unconvincing in this instance!"

I'm not arguing this evidence is conclusive. I am arguing that someone like Gladwell who says "we know" the truth about all this is talking their biases, not science.

Anonymous said...

What sorts of questions do they use in these tests to measure IQ?

The only "IQ tests" I've seen over the years, seem sort of questionable on the surface. For example, many of those Mensa tests look like an exercise in pattern recognition and/or extrapolation. Other tests like the SAT, MCAT, etc ... look sort of like a more rigorous pen-and-paper version of the TV game show Jeopardy or the board game Trivial Pursuit. Some types of questions even look like a rigorous pen-and-paper version of the game "Win, Lose, or Draw" or Pictionary, done in reverse.,_Lose_or_Draw

As to whether our host Steve is genuinely the type of person who is overly proud of his IQ to the point of obsession, as suggested by various commenters attacking him on the tone of his numerous posts on IQ and genetics, I simply don't know since I've never met Steve in person. Though with that being said, I do understand the mindset of people who are overly proud of their IQ scores. My family and extended family are the types who were extremely obsessed about their intelligence and thought that they were the intellectual "chosen ones", albeit in their own minds. (My family are Ashkenazi jews originally from Berlin, who left Germany and moved to England before 1933).

For most of my childhood and youth, my parents were constantly pushing me to achieve "high grades" in things like school, and other intellectual matters like thinly veiled "intelligence tests". It always felt like I was being "cross examined" by my "lawyer" parents everyday. (My parents were lawyers, and my mother's side of the family was also filled with tons of lawyers. It was as if they took home their work with them, and were still "lawyering" when they were at home).

I didn't realize it at the time, but in hindsight my parents were the types who had a deep insecurity about themselves with respect to their intelligence. No matter how big their achievements were and how much other people told them they were smart and intelligent, at their core they still thought they were "stupid" and "unworthy". (Besides their law degrees, my father also had a masters degree in pure math from an Israeli university, and my mother had a degree in pure math and philosophy from an American ivy league university).

Over the years I have met many other highly educated and/or intelligent people with a similar neurosis as my parents. When I got to know some of these people better, many eventually admitted they had a deep general insecurity about themselves which just happened to manifest itself in how they viewed their own intelligence. In other people this type of deep insecurity may manifest itself in other ways besides intelligence, such as body image, personal charisma, their "toys", how big of a "rebel" they are, etc ...

Anonymous said...

You were very lucky to have such parents.

Dave Bacon said...

It's rediculous. It's disgusting.
It's America. FAT and STUPID

Oh, that's just priceless (or is it priseless?)

Anonymous said...

way over my head bacon.

you are a fat person yoursef i believe. and as everyone knows iq has a negative correlation
with girth.

all i said was true bacon.

take presidents vs pms

thatcher v reagan
obama v brown
bush v major
bush v blair
fdr v churchill

the uk elite are smarter and much less pushy.

Dave Bacon said...

Indeed anonymous coward you have got me there! And its so funny, you know the bit where you call me fat and stupid. You Brits just crack me up.

But why, if you are of such amazing IQ, do you hide behind the anonymous name. Shouldn't the world know where such amazing insights come from? And we want pictures of your skinny stomach as well!

Steve your anonymous commenters are so cute.

Anonymous said...

Bacon, I love you.

I'm not a Brit.

I was born in and live still in Portland, Oregon.

Despite my GRE and SAT scores, I'm not so smart (see following).

I love you because you are an accomplished gentile.

Your identity as American, academic, physicist, smart person, etc. All are distractions.

Francis Bacon is and was a more important person than Einstein, but because the English don't control the mass media you'd never know it.

Blog Archive