> Using twenty years of earnings data on Finnish twins, we find that about 40% of the variance of women’s and little more than half of men’s lifetime labour earnings are linked to genetic factors. The contribution of the shared environment is negligible. We show that the result is robust to using alternative definitions of earnings, to adjusting for the role of education, and to measurement errors in the measure of genetic relatedness.
There is extensive literature on twin studies that finds effects of roughly this magnitude on a range of health/earnings/educational metrics. 40%-50% genes 40%-50% random/non-shared environments and small impact of shared environment.
Also IQ seems structured this way, but even more heritable and almost nothing to do with shared environment.
Incidentally IQ is also strongly correlated with lifetime earnings. In fact if you can know only one thing about a job candidate, it's the measure you should pick.
This means the manager needs to correctly align employee incentives. For example pay sales people a commission or give employees shares. This is business school 101.
I do not believe a person's salary history gives you any useful information about their effectiveness as an employee.
I have worked with people who had higher salaries than I did who were clearly much worse at the job (a backend architect at a past gig told me that UTF-8 supported only 256 characters, and refused to acknowledge it when I pointed out sources that showed he was wrong).
Am I correct in reading this as you making the claim that your salary is not directly (or even at all) commensurate with your contribution to the company?
Sorry, I see how that can be confusing that I switched topics like that. I mean IQ is the best single measure predictor of success on the job. It's also strongly correlated with lifetime earnings. And it should surprise nobody that success on the job and lifetime earnings are very tightly coupled.
It has almost nothing to do with shared environment in the wealthy first world countries we study this in. But malnutrition, disease, and other factors can certainly have large effects. The introduction of iodized salt in the US raised IQs by 10 points in certain areas where people's diets and water were sufficiently iodine-deficient.
And the 'free lunch program' raised public health/reduced malnutrition by a large factor when introduced in the US. Urban legend has it that the military mandated it because of poor WWII conscript health, but free lunch predated that by 50 years. It was a federal act only after WWII because of scarcity and the decline of local programs.
Yes, that's an excellent point. The socioeconomic status of your parents is a big influence up to a point. In developed counties we don't see this effect as much because it's the baseline. Almost everybody gets sufficient nutrition and medical care to meet the very low standards where it would otherwise negatively affect your intelligence and development.
So, they do this by comparing correlations in MZ (genetically identical) and DZ (typical sibling genetics) twin outcomes. Doesn't that assume that the fact of being DZ vs MZ does not itself create environmental (non-genetic) effects? The identical twins I know face particular social pressures to respond to their twin identity, which could drive towards outcomes that are more correlated -- or not! But it seems weird to base a branch of literature on the assumption that it's not. Is this addressed?
You can also compare identical twins that grow up in the same family, vs. identical twins that were separated at birth... although I imagine you wouldn't have a lot of samples.
That's a reasonable sort of question but would you say that MZ twins are treated differently enough to explain 40% of variation in earnings? If they were all locked in cellars out of shame I could see that being the case but that doesn't square with the identical twins I've known.
But it actually is addressed by other studies such as adoption studies that show generally the same results. And, of course, there are potential problems with adoption studies too. But wouldn't it be suspicious if every[1] method of studying the problem we tried gives fairly close numbers but only because they all happen to be biased by the same amount in the same direction?
[1]GWASes give lower numbers but we know that there are many sorts of genetic interaction they don't cover so this is expected.
This is an empirical question and as far as we can tell if there’s pressure to be more similar or more different it washes out. Identical twins reared together are no more similar than ones raised apart on a huge host of traits.
Yes that is valid point about the limitation of this approach. There are 3 methods to address it, none of them are perfect:
(1) Statistically compare the population of twins vs the population of non-twins. The paper I linked to has more specifics on this, but twins earn similar amounts of money, live similar lengths of time, do similarly well in exams vs. non-twins, so it any twin-specific genetic effect has to simultaneously be strong enough to cause this discrepancy, but not strong enough to impact population means.
(2) Compare adoptees to genetic offspring of parents. This has the same issue that you pointed out (maybe there is something specific about adoptees) but the two together find roughly similar effect magnitudes, which reenforce each-other. I'm actually a bit more concerned about adoption being less 'clean' an identification strategy than twin studies, because it's a complicated social process.
(3) Look at genetic variation at population level and find genetic markers that explain variation, using Mendelian randomization as identification strategy - this is technically very challenging, and has only recently become feasible. So far research have only been able to explain around 7% of IQ for example with specific genetic markers, but it will be interesting to see if this creeps up as the field matures.
The big puzzle here is the large variation that we between cultures - moving time-periods, or countries can significantly change your life-expectancy, earnings, test-scores or even IQ. I don't know how to reconcile this with the strange finding of this research strand is that family level environmental effects (shared environment) seem to be much less important than the conventional wisdom would suggest.
Wasn't there also a paper that came out fairly recently looking at Swedish administrative data instead which showed something on the order of 1/4 gentic 1/4 shared environment? My general impression has been that lifetime earnings had a shared environmental factor from parental help that wasn't there in other things like scholastic achievement. I'll see if I can track that down. Though possibly the difference might just be high investment income from some children moving the average in studies that take a mean rather than a median.
Strong disagree that IQ is gonna be an explanatory variable here. Not to get political here, but DJTs children, none are “outliers” as far as intelligence goes but all will inherit their fathers (outlier WRT income) income generation capacity within an order of magnitude.
This study found no correlation between environment and earnings when genes were accounted for. So you have a gene part and a random part, but parenting and such doesn't matter.
Of course it might not apply to a few special cases like if you inherited a billion dollars, but for 99% of us that is not the case.
> Actually just said by Donald Trump Jr: "I wish my name was Hunter Biden. I could go abroad and make millions off my father's presidency. I'd be a really rich guy"
You can probably make some reasonable intelligence estimates off that sort of statement.
While IQ isn’t inherently racist, the data is racially biased. If you correlate IQ to race, you may find minority Y has a lower average IQ (true) which may falsely imply that the cause for X is being of race Y.
Discussions about IQ (which isn’t a particularly useful metric anyway, for various reasons) tend to veer into racist territory because people end up associating wealth and success with IQ, and since a disproportionate amount of wealth in America is controlled by whites people, and since minorities are more likely to live in poverty, this causes some people to suggest that those specific minorities have lower IQs or something. Basically, it brings up old racial stereotypes of non-whites not being as intelligent as whites, which is a very old, racist canard.
"because people end up associating wealth and success with IQ, and since a disproportionate amount of wealth in America is controlled by whites people, and since minorities are more likely to live in poverty, this causes some people to suggest that those specific minorities have lower IQs or something"
This is really not how IQ testing is done. Rather than guessing about correlations in the environment to IQ and working backwards to come up with an estimate for IQ, generally IQ tests are administered.
Which is why it's hard to understand how the study suggests environment plays such an insignificant role.
If you're born into a very poor home, and to use your example, a poor black home in the US, I'm not convinced my IQ would overcome my environmental handicap.
How many times out of 10 would you expect to make it to college and earn a high paying job while growing up black in some of the rougher city's and neighborhoods of the US?
Differences between people in a protected class and others are generally considered racist now. Heck, I'm not sure if saying Orthodox Jews tend to have less penile skin is ok.
Downvotes say apparently someone is offended by my factual statement.
> we assumed ... that there is no correlation between genetic
factors and the shared environment
I'm no expert, but that seems like the biggest issue. I mean the study says that income is heritable, which by its very nature would affect the environment that right?
You can estimate true heritability by comparing fraternal and identical twins, both of which will have shared environment, but identical twins are identical genetically and fraternal ones are as similar as any other sibling pair.
> The classical twin design compares the similarity of monozygotic (identical) and dizygotic (fraternal) twins. If identical twins are considerably more similar than fraternal twins (which is found for most traits), this implicates that genes play an important role in these traits. By comparing many hundreds of families with twins, researchers can then understand more about the roles of genetic effects, shared environment, and unique environment in shaping behavior.
> (i.e., within-pair genetic differences are not
correlated with the within-pair environmental differences; see e.g. Stenberg (2013) who
stresses the importance of this assumption for the interpretation of heritability estimates),
and iv)
So this is only about whether sibling differences are correlated with the environment -- which seems much more reasonable. On the other hand, the very next one is problematic:
> [We assume] there is no assortative mating. The last assumption would not hold if the genotypes of the parents were correlated (Posthuma et al. 2003).
Tendency of parents to have some genetic sorting seems both easy to measure and quite likely - why not dig in to its impact?
Finally, another thing that makes the result hard to generalize (emphasis added):
> We show that in the relatively equitable economic and institutional environment of Finland, the share of variance of lifetime earnings explained by education is clearly less than a tenth (in our data).
The authors acknowledge that they were measuring the size of effect in an environment that has far less variation in the variable than many. This seems like a good place to confirm the existence of an effect, but a bad place to be confident of its size, and a worse place from which to generalize about more heterogenous environments. Compare Finland's GINI index (income concentration) to US, UK, or China to get an idea of how much more variation there is in those societies: https://www.indexmundi.com/facts/indicators/SI.POV.GINI/comp...
>"...the share of variance of lifetime earnings explained by education is clearly less than a tenth (in our data). This comparison suggests that the variation in lifetime earnings that can be attributed to genetic variation is not negligible and warrants attention. The results of our auxiliary analyses also suggest - but do not conclusively show -that removing the effects of education on the lifetime earnings of the cohorts we study does not change these heritability estimates."
Unfortunately I can't dedicate too much time to reading the article right now, but did anybody understand how they managed to isolate the genetic vs environmental components?
You do it by comparing identical twins against fraternal twins and see how the variance differ.
That way you have 3 variables, genes (identical twins have same genes, fraternal twins just have similar), the home and parenting part, and all other parts that aren't related to your parents at all. This study in particular found that just the genes and the other factors were relevant, while the home wasn't.
But shouldn't they also include people with completely different genes, but the same educational background? Fraternal twins share less than identical ones, but they still share a lot. It would be interesting to see if unrelated people who come from the same environment exhibit vastly different earnings outcomes.
I assume it is hard to objectively define and measure 'same environment' for arbitrary pairs. Using sibilings (fraternal or twins) is a way around that.
While I think on the whole our society is very biased toward a Boasian view of nature versus nurture: "give me a child and I can raise it to have any personality and aptitudes via deliberate rearing techniques." I'm skeptical of some of the conclusions being drawn from this study.
Environment has not been entirely separated out here it seems. And while a comparison of fraternal and identical twins detangles things somewhat it doesn't completely.
My intuition is that a person's personality and aptitudes (around 50% heritable according to many studies) can help them to take advantage of their environment in certain ways. So it's not that environment doesn't matter at all. It's just that people with varying attributes will leverage their environment in different ways, leading to different outcomes.
It would be even more interesting to me to see the personality profiles of the children compared against their future earnings. My guess is there would be a strong correlation since personality -> interests -> career choices.
Then it would just so happen that identical twins have similar personalities. Leading to the results we see here.
Admittedly I'm hopeful that if we could figure out the correlation between personality and environment maybe we could shift our focus away from the unhealthy extremes of 100% nurture (leads to children being pushed by their parents to fit into a mold that may not suit them) and 100% nature (genetic fatalism leads to apathy and hopelessness). Instead what if we took a child's natural gifts and personality profile into account and tailored their environment to maximize their potential within those constraints? Seems like a more hopeful and useful path.
> While I think on the whole our society is very biased toward a Boasian view of nature versus nurture: "give me a child and I can raise it to have any personality and aptitudes via deliberate rearing techniques." I'm skeptical of some of the conclusions being drawn from this study.
Boas was wrong.
> Genetic Influence on Human
Psychological Traits
> There is now a large body of evidence that supports the conclusion that individual differences in most, if not all, reliably measured psychological traits, normal and abnormal, are substantively influenced by genetic factors. This fact has important implications for research and theory building in psychology, as evidence of genetic influence unleashes a cascade of questions regarding the sources of variance in such traits. A brief list of those questions is provided, and representative findings regarding genetic and environmental influences are presented for the domains of personality, intelligence, psychological interests, psychiatric illnesses, and social attitudes. These findings are consistent with those reported for the traits of other species and for many human physical traits, suggesting that they may represent a general biological phenomenon.
I absolutely don't disagree with you. Personality traits and aptitudes are strongly inherited, along with some genetic randomness and some environmental conditioning. What I don't like is full on genetic determinism. Human beings transmit both genes AND memes for a reason. As sentient beings we inhabit two universes. The one made of atoms and the one made of thought and ideas.
Compare, say, humans and ants.
Ants seem to have a ROM of sorts: software embedded in genes and immutable within a single organism's lifetime. They only inhabit the world of atoms.
Humans have a ROM to cover basic functions but also writable memory. We can change our behaviors within a single lifetime and then if that wasn't cool enough we can also TRANSMIT those behaviors without genes via the world of thought and ideas (language).
We ought to never forget that. Fatalism binds us too strongly to the physical world. It's a bad path that leads to nihilism (struggle against nature is futile), cruelty (everyone deserves their lot in life since if they were capable of more they would have achieved it), and despair (self-actualization is impossible, I am an automaton).
It would be valuable to try and reproduce these findings in a country like the US or UK whose citizens have been found to exhibit much lower socioeconomic mobility [1] whereas the Nordic countries (as well as Canada and Australia) are known to provide much more mobility. I would be surprised if the results held up unaffected.
I suspect that genes that affect immune system play big role in the heritable intelligence just like immune plays role in the environmental factors affecting IQ.
60 comments
[ 5.4 ms ] story [ 107 ms ] threadhttps://www.researchgate.net/profile/Bruce_Sacerdote/publica...
Incidentally IQ is also strongly correlated with lifetime earnings. In fact if you can know only one thing about a job candidate, it's the measure you should pick.
You want someone who will do the work well and will help the company succeed.
Indeed, you could argue that a high earner is, all else being equal, something you should avoid selecting as a hiring manager.
I have worked with people who had higher salaries than I did who were clearly much worse at the job (a backend architect at a past gig told me that UTF-8 supported only 256 characters, and refused to acknowledge it when I pointed out sources that showed he was wrong).
But it actually is addressed by other studies such as adoption studies that show generally the same results. And, of course, there are potential problems with adoption studies too. But wouldn't it be suspicious if every[1] method of studying the problem we tried gives fairly close numbers but only because they all happen to be biased by the same amount in the same direction?
[1]GWASes give lower numbers but we know that there are many sorts of genetic interaction they don't cover so this is expected.
(1) Statistically compare the population of twins vs the population of non-twins. The paper I linked to has more specifics on this, but twins earn similar amounts of money, live similar lengths of time, do similarly well in exams vs. non-twins, so it any twin-specific genetic effect has to simultaneously be strong enough to cause this discrepancy, but not strong enough to impact population means.
(2) Compare adoptees to genetic offspring of parents. This has the same issue that you pointed out (maybe there is something specific about adoptees) but the two together find roughly similar effect magnitudes, which reenforce each-other. I'm actually a bit more concerned about adoption being less 'clean' an identification strategy than twin studies, because it's a complicated social process.
(3) Look at genetic variation at population level and find genetic markers that explain variation, using Mendelian randomization as identification strategy - this is technically very challenging, and has only recently become feasible. So far research have only been able to explain around 7% of IQ for example with specific genetic markers, but it will be interesting to see if this creeps up as the field matures.
The big puzzle here is the large variation that we between cultures - moving time-periods, or countries can significantly change your life-expectancy, earnings, test-scores or even IQ. I don't know how to reconcile this with the strange finding of this research strand is that family level environmental effects (shared environment) seem to be much less important than the conventional wisdom would suggest.
Links https://www.sciencedirect.com/science/article/pii/B008042707... (gated)
https://en.wikipedia.org/wiki/Mendelian_randomization
https://www.newscientist.com/article/2163484-found-more-than...
https://en.wikipedia.org/wiki/Flynn_effect
Of course it might not apply to a few special cases like if you inherited a billion dollars, but for 99% of us that is not the case.
How on earth could you possibly know this?
> Actually just said by Donald Trump Jr: "I wish my name was Hunter Biden. I could go abroad and make millions off my father's presidency. I'd be a really rich guy"
You can probably make some reasonable intelligence estimates off that sort of statement.
Makes complete sense to me.
Not sure what's racist about that though. It's just reality, facts about humans.
This is really not how IQ testing is done. Rather than guessing about correlations in the environment to IQ and working backwards to come up with an estimate for IQ, generally IQ tests are administered.
Raven's Progressive Matrices [1] is an example of a test that doesn't depend upon knowledge of tea service items.
[1] https://en.wikipedia.org/wiki/Raven%27s_Progressive_Matrices
If you're born into a very poor home, and to use your example, a poor black home in the US, I'm not convinced my IQ would overcome my environmental handicap.
How many times out of 10 would you expect to make it to college and earn a high paying job while growing up black in some of the rougher city's and neighborhoods of the US?
Downvotes say apparently someone is offended by my factual statement.
I'm no expert, but that seems like the biggest issue. I mean the study says that income is heritable, which by its very nature would affect the environment that right?
Can you explain this a bit more?
> The classical twin design compares the similarity of monozygotic (identical) and dizygotic (fraternal) twins. If identical twins are considerably more similar than fraternal twins (which is found for most traits), this implicates that genes play an important role in these traits. By comparing many hundreds of families with twins, researchers can then understand more about the roles of genetic effects, shared environment, and unique environment in shaping behavior.
> (i.e., within-pair genetic differences are not correlated with the within-pair environmental differences; see e.g. Stenberg (2013) who stresses the importance of this assumption for the interpretation of heritability estimates), and iv)
So this is only about whether sibling differences are correlated with the environment -- which seems much more reasonable. On the other hand, the very next one is problematic:
> [We assume] there is no assortative mating. The last assumption would not hold if the genotypes of the parents were correlated (Posthuma et al. 2003).
Tendency of parents to have some genetic sorting seems both easy to measure and quite likely - why not dig in to its impact?
Finally, another thing that makes the result hard to generalize (emphasis added):
> We show that in the relatively equitable economic and institutional environment of Finland, the share of variance of lifetime earnings explained by education is clearly less than a tenth (in our data).
The authors acknowledge that they were measuring the size of effect in an environment that has far less variation in the variable than many. This seems like a good place to confirm the existence of an effect, but a bad place to be confident of its size, and a worse place from which to generalize about more heterogenous environments. Compare Finland's GINI index (income concentration) to US, UK, or China to get an idea of how much more variation there is in those societies: https://www.indexmundi.com/facts/indicators/SI.POV.GINI/comp...
edit: but it seems they do not use adoption data.
That way you have 3 variables, genes (identical twins have same genes, fraternal twins just have similar), the home and parenting part, and all other parts that aren't related to your parents at all. This study in particular found that just the genes and the other factors were relevant, while the home wasn't.
Environment has not been entirely separated out here it seems. And while a comparison of fraternal and identical twins detangles things somewhat it doesn't completely.
My intuition is that a person's personality and aptitudes (around 50% heritable according to many studies) can help them to take advantage of their environment in certain ways. So it's not that environment doesn't matter at all. It's just that people with varying attributes will leverage their environment in different ways, leading to different outcomes.
It would be even more interesting to me to see the personality profiles of the children compared against their future earnings. My guess is there would be a strong correlation since personality -> interests -> career choices.
Then it would just so happen that identical twins have similar personalities. Leading to the results we see here.
Admittedly I'm hopeful that if we could figure out the correlation between personality and environment maybe we could shift our focus away from the unhealthy extremes of 100% nurture (leads to children being pushed by their parents to fit into a mold that may not suit them) and 100% nature (genetic fatalism leads to apathy and hopelessness). Instead what if we took a child's natural gifts and personality profile into account and tailored their environment to maximize their potential within those constraints? Seems like a more hopeful and useful path.
Boas was wrong.
> Genetic Influence on Human Psychological Traits
> There is now a large body of evidence that supports the conclusion that individual differences in most, if not all, reliably measured psychological traits, normal and abnormal, are substantively influenced by genetic factors. This fact has important implications for research and theory building in psychology, as evidence of genetic influence unleashes a cascade of questions regarding the sources of variance in such traits. A brief list of those questions is provided, and representative findings regarding genetic and environmental influences are presented for the domains of personality, intelligence, psychological interests, psychiatric illnesses, and social attitudes. These findings are consistent with those reported for the traits of other species and for many human physical traits, suggesting that they may represent a general biological phenomenon.
http://humancond.org/_media/papers/bouchard04_genetic_influe...
Compare, say, humans and ants.
Ants seem to have a ROM of sorts: software embedded in genes and immutable within a single organism's lifetime. They only inhabit the world of atoms.
Humans have a ROM to cover basic functions but also writable memory. We can change our behaviors within a single lifetime and then if that wasn't cool enough we can also TRANSMIT those behaviors without genes via the world of thought and ideas (language).
We ought to never forget that. Fatalism binds us too strongly to the physical world. It's a bad path that leads to nihilism (struggle against nature is futile), cruelty (everyone deserves their lot in life since if they were capable of more they would have achieved it), and despair (self-actualization is impossible, I am an automaton).
[1] https://www.oecd.org/centrodemexico/medios/44582910.pdf
What does this mean?