Parenting matters a lot in areas that are not being measured by psyshometricians. Religion, table manners, personal hygiene, etc. are traits that are transmitted mainly through nurture. Yes, the first law of behavior genetics says that very trait is heritable, including religiosity. But parents are the ones that teach their kids to brush their teeth after every meal. Not all cultures teach this.
And ultimately these behaviors that are taught at home are what defines a culture.
Grooming is much more influenced by peers and culture, not parents. If most kids followed their parents' advice on grooming past puberty (which virtually none of them do, and often get into major arguments with their parents about that very issue), they'd all look like huge dorks.
Take a typical secular ashkenazi jew and a typical chassidic jew. If you look closely, you can see underlying similarities, but at a prety obvious level they are very different. Some of the nature-only polemic boils down to saying these differences aren't important because tests haven't been devised to measure them.
Then you compare Ashkenazi jews with other Ashkenazi jews and Chassidic jews with other Chassidic jews, they're obviously very different within the groups.
What? I'm talking about having all the things that are associated with a good life but still being miserable because of personality or temperament. A rare muscle dysfunction that interferes with your life is not an objectively good life and it's certainly not the majority of those cases.
Parenting can make a BIG difference in choices made. My son was probably more criminally inclined than I. I had figured out at a teenager that it if your objective was to live a free life, it was easier to be a legitimate business than a criminal business (even if nobody is complaining about your activities, there is always the IRS asking questions about your income). So I would watch, monitor, and call him on activities, and ask him how he intended to not get caught. Eventually he realized that he did not have the detailed planning skills to cover his traces and that he was better off being legitimate. I pointed out to him that being able to think like an attacker was critical to defensive actors, who are both legitimate and can be paid very well. I also pointed out that he would have to undergo routine background checks and would have to limit his activities and associates accordingly. He did his MIS in data security and works successfully in the field.
I think this post is a helpful clarification on the matter, thank you.
There is but one thing that wasn't discussed, which is that parenting may not matter much on aggregate, on a societal level, but it does not mean it cannot matter heavily if you're on one or the other end of the parenting spectrum.
Guarantee that if you slap your kid around all day it will have an effect beyond you transmitting your child-beating genetics.
On the other hand I absolutely refuse to believe you could not have a strong effect if you chose to devote yourself to your children's upgringing (without being overbearing in their teens).
A minor quibble on wording in the preface. "Statistical insignificance" of course never means zero (but rather indistinguishable from zero), just as statistical significance does not always mean meaningful. If I understand correctly, noisy data is as great an impediment to statistical significance as power.
The comments are interesting as so many would deny, minimize or modulate the effects you describe.
One thing I'm not understanding about the square root thing - so sure, heritability is a variance measure, and I understand the argument.
But if you took a regular distribution, something like:
Heritability: 80%
Shared Env: 5%
Unshared Env: 10%
Error: 5%
I understand if you want to compare the 80% vs 5%, you would take the square root to get better relative effect sizes. So heritability is "only" 4x Shared Env at 5%. But this doesn't scale to explain the "whole" effect size, or think about things holistically. If you take the square root of each of them, they add to more than 1 (in this case, 1.657).
When we think about a Shapley-value-esque entire partioning of explaining the relative importance of each of these factors, I think you should still think in terms of the base variance-explanation statistic - the 80% vs 5%.
After all, saying heritability is "only" 4x as strong as Shared Env and Error in this case, and "only" 2.8x as strong as Unshared Env, doesn't really give you intuition on the whole partioning of explanatory power.
Unless I've misunderstood, I'm actually confused on this point and would appreciate pointers from people who have spent more than 10 min thinking about this.
When trying to determine how much variation is attributable to the common environment and how much to variation in genes, then you just take h^2 and c^2 directly.
But, when trying to predict the average value of Y from X, you just want the correlation coefficient between X and Y, not variance explained.
For example: Say r between X and Y is .4, whereas r between W and Y is .2. that implies there is twice as strong an effect on Y from X than the effect on W from Y. But the variance explained between X and Y is .16, which is 4 times the variance explained between W and Y.
Heritabilty h^2 and Common Environment c^2 are each variance explained; hence the suggestion of taking the square root.
Regarding Robert Plomin’s “Parents matter, but they don’t make a difference”, I wonder if we should interpret that as when we hear that height in basketball, say, or intelligence in law practice, are not that relevant in how performance varies in those occupations.
Height doesn’t matter as much as one thinks, as long as one is talking of professional basketball players, who are way taller than average; and intelligence may not matter as much as one thinks in the legal profession because most lawyers are more intelligent than average to begin with.
Likewise, parents don’t make much of a difference not because all their efforts have a negligible impact, but because most parents in stable, functional families (and even in not-so-stable ones) make no small efforts to ensure their children's well-being.
Plomin does often focus on, e.g., psychological traits like intelligence, and he does occasionally make caveats that parents matter for certain outcomes, even for example educational attainment. But I am not so much trying to suggest that he is wrong about the facts, but rather that I disagree with making such statements, even if we did fully agree about the underlying facts.
To me, if parents make a difference for social outcomes that we consider hugely important in our lives (educational attainment, income, crime, etc), then it's not sensible make statements like this, even if he acknowledges it matters in those circumstances. It's like saying "they don't make a difference... well, except they do make an important difference here, there and there." At the very least, I don't think that's a good way to communicate and accurately describe reality.
All in all, it was never my goal to "debunk" Plomin, but instead just make the positive case that nurture matters (to some extent) to any skeptics out there.
> Yes, the adage that both nature and nurture matter is boring. It is also true.
No, it does not appear to be true regarding who we are. It terms of what happens to us, what we believe, or some things we do, fine. Net worth, religion, education, all of those things show effects from the common environment. But when it comes to intelligence and personality, you will be who you are: https://thingstoread.substack.com/p/you-will-be-who-you-are
My previous understanding was that lifetime income showed no appreciable effects from shared environment, so Bird's review here surprised me somewhat. It's a good article in any case.
I am curious as to what 'cognitive skills' refers to here, as something distinct from IQ?
Parenting matters while they are still living with you, but declines rapidly after they leave.. It doesn't go to zero, but thirty, fifty years out there's not much left. Genetics has taken over by then. Parents of young children believe parenting matters but cannot make the case from their own experience. It is the parents of highschoolers who make the strongest emotional case and who are the hardest to convince that parenting doesn't matter, because they can see the fruits of almost two decades of parenting right before their eyes, at the time that it is most visible. Yes, there is convenient reinterpretation and seeing what one wants to see, but a lot of it is also true. For a while.
The real question is if the difference they make can be predicted and if it's even relevant to be noticed in an individiual's life instead of just small statistical differences.
Isn’t it possible that parents have large effects, but the effects from different parents often go in opposite directions, thereby largely cancelling out?
For example, suppose 5% of kids have a deadly allergy to peanuts, and 5% to tree nuts. Half the parents believe in feeding peanuts, and the other half tree nuts. Then, which parent you have can matter a lot (life or death), but it won’t show up statistically.
That *could* be true, but every so often you'd also get *both* parents believing in feeding their child peanuts, which would have double the effect size(?), and should statistically show up as a shared environmental component for child mortality.
Parenting matters a lot in areas that are not being measured by psyshometricians. Religion, table manners, personal hygiene, etc. are traits that are transmitted mainly through nurture. Yes, the first law of behavior genetics says that very trait is heritable, including religiosity. But parents are the ones that teach their kids to brush their teeth after every meal. Not all cultures teach this.
And ultimately these behaviors that are taught at home are what defines a culture.
Grooming is much more influenced by peers and culture, not parents. If most kids followed their parents' advice on grooming past puberty (which virtually none of them do, and often get into major arguments with their parents about that very issue), they'd all look like huge dorks.
"Religion, table manners, personal hygiene, etc. are traits that are transmitted mainly through nurture. "
Evidence? From all you know they simply inherit what their parents will do in a given culture.
Take a typical secular ashkenazi jew and a typical chassidic jew. If you look closely, you can see underlying similarities, but at a prety obvious level they are very different. Some of the nature-only polemic boils down to saying these differences aren't important because tests haven't been devised to measure them.
Then you compare Ashkenazi jews with other Ashkenazi jews and Chassidic jews with other Chassidic jews, they're obviously very different within the groups.
One of the simplest things parents control is whether their kids have a happy childhood.
Many kids have what are objectively good lives, and even good relations with their parents, but they are still miserable.
It's much more likely to be Bowlbyan attachment issues than psoas muscle issues.
Farming infants out to daycare / early childhood "education", a notoriously high-turnover industry, is psychologically damaging them.
In the vast majority of those cases, it's psoas muscle dysfunction. That's generally what causes "objectively good life, but still miserable"
What? I'm talking about having all the things that are associated with a good life but still being miserable because of personality or temperament. A rare muscle dysfunction that interferes with your life is not an objectively good life and it's certainly not the majority of those cases.
Having an abnormally tight/restricted psoas (prob ~50% of the population) will have quite a negative impact on your temperament
https://twitter.com/nightfire0/status/1709362360190919001
Parenting can make a BIG difference in choices made. My son was probably more criminally inclined than I. I had figured out at a teenager that it if your objective was to live a free life, it was easier to be a legitimate business than a criminal business (even if nobody is complaining about your activities, there is always the IRS asking questions about your income). So I would watch, monitor, and call him on activities, and ask him how he intended to not get caught. Eventually he realized that he did not have the detailed planning skills to cover his traces and that he was better off being legitimate. I pointed out to him that being able to think like an attacker was critical to defensive actors, who are both legitimate and can be paid very well. I also pointed out that he would have to undergo routine background checks and would have to limit his activities and associates accordingly. He did his MIS in data security and works successfully in the field.
I think this post is a helpful clarification on the matter, thank you.
There is but one thing that wasn't discussed, which is that parenting may not matter much on aggregate, on a societal level, but it does not mean it cannot matter heavily if you're on one or the other end of the parenting spectrum.
Guarantee that if you slap your kid around all day it will have an effect beyond you transmitting your child-beating genetics.
On the other hand I absolutely refuse to believe you could not have a strong effect if you chose to devote yourself to your children's upgringing (without being overbearing in their teens).
Parenting is vital, but the minimum level is lower than many people seem to think.
It is always possible to stunt a child.
does shared environment include social interventions(schoolling, wars, revolutions etc.) ?
Excellent work!
Informative, TY.
A minor quibble on wording in the preface. "Statistical insignificance" of course never means zero (but rather indistinguishable from zero), just as statistical significance does not always mean meaningful. If I understand correctly, noisy data is as great an impediment to statistical significance as power.
The comments are interesting as so many would deny, minimize or modulate the effects you describe.
One thing I'm not understanding about the square root thing - so sure, heritability is a variance measure, and I understand the argument.
But if you took a regular distribution, something like:
Heritability: 80%
Shared Env: 5%
Unshared Env: 10%
Error: 5%
I understand if you want to compare the 80% vs 5%, you would take the square root to get better relative effect sizes. So heritability is "only" 4x Shared Env at 5%. But this doesn't scale to explain the "whole" effect size, or think about things holistically. If you take the square root of each of them, they add to more than 1 (in this case, 1.657).
When we think about a Shapley-value-esque entire partioning of explaining the relative importance of each of these factors, I think you should still think in terms of the base variance-explanation statistic - the 80% vs 5%.
After all, saying heritability is "only" 4x as strong as Shared Env and Error in this case, and "only" 2.8x as strong as Unshared Env, doesn't really give you intuition on the whole partioning of explanatory power.
Unless I've misunderstood, I'm actually confused on this point and would appreciate pointers from people who have spent more than 10 min thinking about this.
When trying to determine how much variation is attributable to the common environment and how much to variation in genes, then you just take h^2 and c^2 directly.
But, when trying to predict the average value of Y from X, you just want the correlation coefficient between X and Y, not variance explained.
For example: Say r between X and Y is .4, whereas r between W and Y is .2. that implies there is twice as strong an effect on Y from X than the effect on W from Y. But the variance explained between X and Y is .16, which is 4 times the variance explained between W and Y.
Heritabilty h^2 and Common Environment c^2 are each variance explained; hence the suggestion of taking the square root.
Sibling comparison studies also show some role for shared environmental effects(https://easthunter.substack.com/p/those-sibling-comparison-studies)
Regarding Robert Plomin’s “Parents matter, but they don’t make a difference”, I wonder if we should interpret that as when we hear that height in basketball, say, or intelligence in law practice, are not that relevant in how performance varies in those occupations.
Height doesn’t matter as much as one thinks, as long as one is talking of professional basketball players, who are way taller than average; and intelligence may not matter as much as one thinks in the legal profession because most lawyers are more intelligent than average to begin with.
Likewise, parents don’t make much of a difference not because all their efforts have a negligible impact, but because most parents in stable, functional families (and even in not-so-stable ones) make no small efforts to ensure their children's well-being.
Plomin does often focus on, e.g., psychological traits like intelligence, and he does occasionally make caveats that parents matter for certain outcomes, even for example educational attainment. But I am not so much trying to suggest that he is wrong about the facts, but rather that I disagree with making such statements, even if we did fully agree about the underlying facts.
To me, if parents make a difference for social outcomes that we consider hugely important in our lives (educational attainment, income, crime, etc), then it's not sensible make statements like this, even if he acknowledges it matters in those circumstances. It's like saying "they don't make a difference... well, except they do make an important difference here, there and there." At the very least, I don't think that's a good way to communicate and accurately describe reality.
All in all, it was never my goal to "debunk" Plomin, but instead just make the positive case that nurture matters (to some extent) to any skeptics out there.
> Yes, the adage that both nature and nurture matter is boring. It is also true.
No, it does not appear to be true regarding who we are. It terms of what happens to us, what we believe, or some things we do, fine. Net worth, religion, education, all of those things show effects from the common environment. But when it comes to intelligence and personality, you will be who you are: https://thingstoread.substack.com/p/you-will-be-who-you-are
My previous understanding was that lifetime income showed no appreciable effects from shared environment, so Bird's review here surprised me somewhat. It's a good article in any case.
I am curious as to what 'cognitive skills' refers to here, as something distinct from IQ?
"Cognitive skills" is basically an IQ test, measured approximately at age 18 (it's from army conscription data).
Parenting matters while they are still living with you, but declines rapidly after they leave.. It doesn't go to zero, but thirty, fifty years out there's not much left. Genetics has taken over by then. Parents of young children believe parenting matters but cannot make the case from their own experience. It is the parents of highschoolers who make the strongest emotional case and who are the hardest to convince that parenting doesn't matter, because they can see the fruits of almost two decades of parenting right before their eyes, at the time that it is most visible. Yes, there is convenient reinterpretation and seeing what one wants to see, but a lot of it is also true. For a while.
What evidence would you point at suggesting large wilson effects with respect to the traits discussed here?
Debunked.
https://www.josephbronski.com/p/do-parents-make-a-difference
"Parents matter and they make a difference"
The real question is if the difference they make can be predicted and if it's even relevant to be noticed in an individiual's life instead of just small statistical differences.
Isn’t it possible that parents have large effects, but the effects from different parents often go in opposite directions, thereby largely cancelling out?
For example, suppose 5% of kids have a deadly allergy to peanuts, and 5% to tree nuts. Half the parents believe in feeding peanuts, and the other half tree nuts. Then, which parent you have can matter a lot (life or death), but it won’t show up statistically.
That *could* be true, but every so often you'd also get *both* parents believing in feeding their child peanuts, which would have double the effect size(?), and should statistically show up as a shared environmental component for child mortality.