Text

Physicist, Startup Founder, Blogger, Dad

Saturday, March 29, 2014

The truth about social mobility



See also Not (all) in our genes?

The Truth About Social Mobility

Full audio (55 min including Q&A; video is only 22 min highlights)

Many people assume that it is much easier to move between social classes today than at any point in humankind.

However, new research from Gregory Clark, professor of economics at the University of California, Davis, reveals that mobility rates are lower than conventionally estimated and surprisingly resistant to social policies.

By tracking family names over generations to measure social mobility across periods and countries, Clark reveals that more than ever, the only sure route to success is to be born to the right parents. And so we need to come up with new ways to tackle the entrenched force of inherited advantage and avoid creating winner-take-all societies.

Speaker: Gregory Clark, professor of economics, University of California, Davis

37 comments:

Butch said...

Atleast now(I hope) a better future is in sight. The NPTN gene(based on the allele) accounts for 0.5% of iq variation. That is just one gene. We have the technology and money to truly help humanity. All we need now is the mindset to accept it. Once it starts though... it won't stop. If you offer southern europeans the chance to become northern european- within a generation- a whole populus could change. It will be a domino effect. What the vaccine is to disease, this technology will be to poverty. This has to happen.

chartreuse1737 said...

very odd. didn't clark take the opposite tack in "a farewell to alms"? is this the truth finally hitting an economist in the face?

chartreuse1737 said...

you're making the huge assumption that "social competence" is the same thing in every society past and present with the same + and - alleles, and that "social competence" is a virtue.

Iamexpert said...

What about economic mobility in the animal kingdom? Animals can inherit wealth from their parents too. If your mom was the fastest cheetah in the pack, you get to feast on the carcasses she caught. That's survival of the fittest, or survival of the fittest's children (who presumably inherit not only their mom's economic superiority, but the genetic superiority that caused it) . But the difference is food decomposes and thus can't be passed down to grandchildren, let alone great grandchildren, so rich animals don't regress to the mean as rapidly as rich people. If they did, evolution would be impeded.

LondonYoung said...

And even if food lasted, animals do not institute governments to enforce a concept of property ownership - so the "fittest" animals can just seize assets as they please. I presume that the underlying thought in the crusade against the 1% is that either governments more evenly divide the goods, or the 99% of animals will lose tolerance for property law. "The French aristocracy didn't see it coming either", etc...

chartreuse1737 said...

only in america would someone so stupid be in the 1%.

Iamexpert said...

Couldn't one argue that chimps institute a primitive form of government to enforce property ownership? They form hierarchies and I believe the higher ranked males have access to more food and females and the entire troop enforce property rights by killing other troops that come to close to their territory.

Abruzzi_spur said...

said the spartans to the helots. said the bolvian silver mine owner to his miners.

w/o labor property is useless. unless it's personal property. genuine democracy is a way of bargaining between capital and labor. the us is and has always been a dictatorship of the bourgeoisie. it is the most corrupt country in the developed world by far.

Cornelius said...

...reveals that mobility rates are lower than conventionally estimated and surprisingly resistant to social policies.



By tracking family names over generations to measure social mobility
across periods and countries, Clark reveals that more than ever, the
only sure route to success is to be born to the right parents. And so we
need to come up with new ways to tackle the entrenched force of
inherited advantage and avoid creating winner-take-all societies.


As societies reduce the variation in environment/nurture faced by different individuals, then the relative influence of nature must rise. If genes determine most of the attributes that enable success, then you would expect social mobility to decrease in such a society. An increase in assortive mating would tend to exaggerate this trend further.

These are the outcomes we expect in a meritocracy. The West has gone further in approaching meritocracy than any previous society.

LondonYoung said...

interesting, but don't know enough about chimps ...

Kennon Gilson said...

Thoughts:
The study conflates totalitarian and relatively free societies with respectively poor and fairly good data so is misleading from the start. Which may be the intent.
Most 'social' policies, because they're coercive, backfire according to some economists and libertarian thinkers. Yet the solution is more silly studies to expand the flawed policies?

Iamexpert said...

But the rich inherit better environments as well as better genes so they're advantaged regardless of whether nature or nurture dominates.

Abruzzi_spur said...

"The West has gone further in approaching meritocracy than any previous society."


by the west, i suppose "cornelius" doesn't mean the west of only fifty years ago, which was MUCH more meritocractic the west is today. cornelius needs to up his verbal iq. so far he's just stringing words together.

StatCruxDumVolviturOrbis said...

no need to respond except in jest. cornelius has been preferred by the LEAST meritocratic country in the developed world, perhaps the entire world, if "meritocracy" is accepted as it is by wikipedia http://en.wikipedia.org/wiki/Meritocracy.

and, obviosuly, the very attachment to "meritocracy" as an absolute rather than a local and particular system of preferment indicates a lack of subtlety.

alfredostroessner said...

never mind. if you're more mathematically sophisticated than steve he won't post you. if you've got a higher gre than steve he won't post you.

alfredostroessner said...

when steve's butt budy paul young says, in so many words, that the poor deserve their poverty, steve posts it. the butt sex is too good even if steve disagrees.

5371 said...

Hilarious nonsense.

5371 said...

Your last sentence is a confession of faith, not an empirically testable proposition.

a last a loved a long the said...

There are two possible explanations other than connections:

1. Cryptic social stratification. There is endogamy within a large group of high ability.

2. With an h^2 of .7, if one twin has an iq of 160, the likelihood the other has an iq of 160 or above is theoretically only 1.4%, and this not taking into account spearman's law of diminishing returns, that the heritability is all in g. That is, if a rich kid and a poor kid have the same high "genetic true score", the rich kid is more likely to have a very high full score.

Iamexpert said...

I strongly suspect that Spearman's law of diminishing returns is a statistical artifact, probably caused by ceiling bumping on certain tests.

a last a loved a long the said...

Razib has a post on Minnesota twins correlation at average age 41. It was < .7 for the WAIS and < .6 for the Raven's. Correction for reliability over a short time period, like a day, is so close to 100% it would make no difference.

So called "additive heritability" may be an example of scientist as Procrustes. One can find an additive effect simply by fitting a hyperplane. But the surface so approximated may be very unlike a plane.

Iamexpert said...

Those numbers make no sense. Jensen reported that IQ has a heritability of 0.8 by later maturity, which would imply that identical twins reared apart should correlate around 0.9 by later adulthood (though how late I'm not sure).

a last a loved a long the said...

They make perfect sense. Heritabilities for IQ vary from 0 to Jensen's (the highest), and all are correct, because they are all heritabilities for different populations.

a last a loved a long the said...

Your explanation would actually reverse the law. A lot of ceiling scores and the variability of scores decreases and are thus more explained by g.


On the Wechsler if one scores +3 sigma on all the subtests one's score isn't 145. It's a lot higher. But one's scores would be entirely explained by g.

Iamexpert said...

Well if the heritability is 0.8, then it would imply the correlation between identical twins reared apart would have to be 0.9 (the square root of 0.8). And yes, I've read the IQ's of identical twins do rise and fall together, showing the same genetically preordained cognitive growth spurts, delays and plateaus.

Jensen reports that the broad heritability of IQ is about 0.45 in children, 0.65 in adolescents, and approaches 0.8 in later maturity. So child twins reared apart should correlate 0.67, teen twins 0.81, and mature twins, nearly 0.9.

a last a loved a long the said...

Well if the heritability is 0.8, then it would imply the correlation between identical twins reared apart would have to be 0.9 (the square root of 0.8).


You can have a plenary indulgence for thinking that, but because the score of both twins in the model is G + E and not a naked G, it turns out that the twin-twin correlation = h^2.

Iamexpert said...

My apologies, I appear to be very confused. If heritability is defined as the percent of variation explained by genes, then the square root of heritability is the phenotype-genotype correlation, which I foolishly equated with the twin-twin correlation. But as you correctly point out, each twin has her own environment, so the correlation between twins would actually be genotype-phenotype correlation squared. Thanks for clearing that up.

a last a loved a long the said...

I don't know why but the heritability, fraction of variance explained, is usually symbolized as h^2 not just h.


So the model is trait = hG + (1-h^2)^(1/2)E.


Needless to say this model is ridiculous.

Butch said...

Statement retracted... thanks for the wakeup.

Iamexpert said...

Well anytime you square a correlation, you get the proportion of variance explained. For example, squaring the 0.4 correlation between IQ and brain size gives 0.16, so 16% of the variation in IQ is explained by brain size.

Thus I can only assume that h is the genotype-phenotype correlation which would make h^2 the proportion of variation in phenotype that is explained by genotype.

a last a loved a long the said...

You get the proportion of variance explained by what though? Think about that.


In the case of twins the correlation squared gives the % of variance explained by the other twin's full G + E score, but not the % of variance explained just by his G score. This latter is just the correlation!

Iamexpert said...

Steve probably has a much better understanding but here's how I explain it to myself (I could be very confused):

Since the twins are reared apart, its assumed they can only correlate for genetic reasons (ignoring prenatal environment). Since correlations are products of factor loadings, the genetic loading in environment A multiplied by the genetic loading in environment B gives the correlation between twins reared in A and B. Thus by taking the square root of the twin-twin correlation, you get the average genetic loading of IQ in environments A and B. Then you square that average genetic loading (h) to get the proportion of variance genes explain (h^2).

a last a loved a long the said...

It's just math. No need to get hung up on twins or genes or environment or iq or whatever.


Just go through the formula for correlation of two variables where one is X + error 1 and the other is X + error 2. Then if you can find variance(X)/(variance(X+error)). And it turns out it's just the correlation of those two variables.

a last a loved a long the said...

I remembered correctly.

In the present study, 30 normal adults were administered both the WPT
and the WAIS on 2 occasions 5 yrs apart. Ss were 17–69 yrs of age at the
2nd testing. Test–retest reliability was .94 for the WPT and .96 for
the WAIS FSIQ.
http://psycnet.apa.org/journals/ccp/51/2/316/

So over a short period reliability would be near perfect for the WAIS. And that's the reliability which should be used to correct the twin-twin correlation if they're give the test at the same time. If one wants to correct with a longer term reliability the correlation he must correct is that of one twin's score at time 2 with the other twin's score at time 1.

So, contra what some claim, the correction for reliability makes no difference for the WAIS. In the Minnesota study MZAs height correlation was .86, which is a lot higher than the almost 100% short term reliable WAIS. The height IQ analogy is bs.

Iamexpert said...

But even 0.69 is a high correlation. It's about the correlation you would expect if the same individuals took the same IQ test in childhood and adulthood, or if the same individuals took the WAIS and then the Raven on the same day.

a last a loved a long the said...

If one wants to correct with a longer term reliability the correlation
he must correct is that of one twin's score at time 2 with the other
twin's score at time 1.


The MZT correlation was .88!

.69 is high. So what? As Steve Jones said, "If everyone smoked, lung cancer
would be a genetic disease." There are ways a particular society is
which are the same for everyone and are, for some, like smoking is for
most. That is, deleterious to their intellectual development.

A redwood withers in the Sonoran Desert, and a saguaro rots in Big Sur.

Ask yourself this question---is it possible my society is less than optimal
for me and is closer to optimal for some others?
What difference does being raised apart make when everyone is raised in
the same "ecosystem"? Why shouldn't it be that some are redwoods in the
Sanoran and others saguaros? And there are examples of just this.

Behavioral genetics is a side show of ideology. It doesn't have the funds or the human capital to do the job right.

Kennon Gilson said...

I would add that social mobility is not the same as ease of movement. The study conflates them. You could have maximum ease of movement but no one moves as lacking what is needed.

Blog Archive

Labels