Do children brought up in the same family share the same environmental influences?

Maybe not.  Here is a fascinating 2011 paper (pdf) from Robert Plomin and Denise Daniels, and here is part of the abstract:

…environmental differences between children in the same family (called ‘‘nonshared environment’’) represent the major source of environmental variance for personality, psychopathology, and cognitive abilities. One example of the evidence that supports this conclusion involves correlations for pairs of adopted children reared in the same family from early in life. Because these children share family environment but not heredity, their correlation directly estimates the importance of shared family environment. For most psychological characteristics, correlations for adoptive ‘‘siblings’’ hover near zero, which implies that the relevant environmental influences are not shared by children in the same family. Although it has been thought that cognitive abilities represent an exception to this rule, recent data suggest that environmental variance that affects IQ is also of the nonshared variety after adolescence.

My favorite part of the paper is the section which discusses how siblings represents a non-shared environment for each other.  For instance my sister grew up with a slightly younger brother and I grew up with…a slightly older sister, which is a somewhat different proposition.

For the pointer I thank Michelle Dawson.


The paper was originally published in 1987. It was reprinted in 2011. Unfortunately, the significance of these findings has never been acknowledged in much of social science, and lots of researchers continue to think that the shared childhood family environment is the major source of outcome differences in adults. These results were an "explosion without a bang", as Neven Sesardic put it, because they were largely ignored despite their importance. Emblematically, Tyler appears to have never heard of them before.

Judith Rich Harris's The Nurture Assumption is a popular treatment of the topic.

Actually, it's a 1987 paper. The link you provided cuts off the 2011 commentary the paper by one of its authors. The commentary summarizes progress on the topic since then. I think many MR readers would benefit from reading it:

Balderdash. How do the authors measure "shared environment"? From this: "twin studies agree: The average IQ cor-relation in over 30 studies is .85 for identical twins and .58 for fraternal twins 28, which suggests again that about 30% of the variance of IQ scores can be accounted for by shared environment." So 85-58 = 27 = about 30%. This means there is 30% more "environment" factors in fraternal twins than identical twins. But though logically sound, I am not sure if IQ is indeed not a function of environment. Let us do a thought experiment: in a poor ghetto, two identical twins are given IQ tests, they score very low. Then fraternal twins are given the same test, and they score equally low. Random adopted pairs of kids are given the same test, and they score the same low scores. Conclusion: there's no genetic component to IQ. But in a rich environment, you get 0.85, 0.58, 0.50 correlation between the pairs--since now the kids have enough vitamins to think straight--so now you can say environment is 15%, 42% and 50% for these pairs. This is just a fanciful example but it shows how these tests can be biased so that one factor influences the other.

That's not the way to calculate these things. These days, SEM model fitting is used to select models that best explain twin correlations, but the simple, traditional approach is that heritability is twice the difference between MZ and DZ correlations, while the shared environment is the difference between the DZ correlation and .5 times heritability.

Therefore, if MZ and DZ correlations are .85 and .58, heritability is 2(.85-.58)=.54, and shared environment is .57-0.54/2=.3. The rest (1-.54-.3=.16) is unshared environment. However, these figures are mostly from studies on children. In adults heritability is higher and shared environment lower.

Let us do a thought experiment

I'm not sure if I understand your thought experiment, but the effects of genes and environment may of course vary within a population. Heritability is the average effect, the proportion of IQ variance explained by heredity across all twin pairs. The relative importance of genes and environment is not necessarily the same across social classes.

"In adults heritability is higher and shared environment lower."

Reversion to the gene.

Your genes are like Bill Belichick. Your parents are like Pete Carrol.

> Your genes are like Bill Belichick

So, consistently excellent and maybe a little bit evil? That explains a few things...

Well, I was trying to think of the human version of the Terminator that would be relatable.

Seems that you, P, are the ideal math lab rat, plugging and chugging variables into a software package that you understand at the operator level.

"I’m not sure if I understand your thought experiment . My thought experiment was the one thought provoking part of the post. The larger point, being that there is no independent or dependent variables, but both depend on each other. The rebuttal to my thought experiment would be that contrary to my assumption, in the ghetto you do see correlation between genetically similar twins, even though they have poor nutrition or environment, that is, you do have pairs of twins that are geniuses or idiots, and their IQ is not random (albeit due to their environment their IQ levels are probably lower than the rich twins). That would rebut my thought experiment.

"The ghetto" is not even close to a bad enough environment to wipe away the effect of genes.

This means there is 30% more “environment” factors in fraternal twins than identical twins.

The effect of environment is the same regardless of twin type. That's a basic assumption in twin studies. MZ twins are more similar only because they are genetically identical. For example, if the average MZ difference is 6 IQ points and the average DZ difference 12 points, MZ twins are 6 points more similar because they are genetically identical, while the remaining 6 points represent non-shared environmental effects which influence MZ and DZ twins equally.

Why do people make such unnecessary assumptions?

The equal environments assumption, which stipulates that the trait-relevant environment of MZ twins is no more similar than that of DZ twins, is not unnecessary. The twin method does not work if the assumption is false. However, it has been extensively tested and found to hold in practice.

The effect of environment is the same regardless of twin type The effects are the same but the levels are not. If "genes" + "environment" = 1, my point is that "environment" = 1-"genes", and therefore, since "genes" for identical twins is greater than "genes" for fraternal twins, the "environment" absolute value is greater for fraternal twins than identical twins. Whether this point has context in the debate above is another matter, as this is not my field and I thank God for that.

I guess my identical twin raised in China will never learn Chinese ;)

I hate this debate because:
1. What is going on is pretty intuitive (yes, identical twins can learn completely different languages, so environment is obviously important) it's just hard to prove it and describe it with elegant math.
2. It seems so reductive and political.
3. The combination of 1 and 2 make discussions insufferable.

Bottom line, I think is, as a parent, all I need to know is where to allocate my attention.

Genes + environment do not equal 1.


What are these acronyms?

Monozygotic and dizygotic. Identical twins and fraternal twins, respectively.

Your thought experiment doesn't make sense, because the absolute scores are meaningless in a correlation. It is not as if the correlation is increased because their scores increased in an abosolute sense.

Let us do a thought experiment: in a poor ghetto, two identical twins are given IQ tests, they score very low.

We should define poor ghetto. Is that in India, Haiti or some part Chicago/DC/NYC/Boston.

I know two identical twins. They put great weight upon their differences (e.g., one has 20-22 eyesight, the other 20-24) and have very disparate ambitions because each has noticed that he's not quite as good as the other at certain things. Thus, one is studying to be an actor, the other to be an engineer.

If they had been raised apart, they might be more similar to each other because they wouldn't be constantly trying to differentiate themselves from each other.


I'm quite familiar with identical twins, and I observe this phenomena all the time. If one likes yellow, the other cannot like yellow.

Also, many identical twins have been through various intrauterine processes, like TTTS, that tend to cause their outcomes to diverge.

As far as I can tell this causes twin studies to systematically underestimate the genetic component of things.

You can see this phenomenon with twins in basketball. Growing up together, one plays the natural position for the pair, the other plays a different position. For instance, Horace Grant, who won three NBA titles with the early 1990s Chicago Bulls, was built like a natural shooting forward, but his slightly more willful brother Harvey got to be the quick forward on their high school team, so skinny Horace learned to play power forward. Horace wound up with the more successful career than Harvey, perhaps because he had worked to be useful outside his comfort zone, perhaps because he was slightly more willing than his identical twin to subordinate himself for the good of the team. (Harvey Grant would not have gotten along with Michael Jordan.)

Judith Rich Harris. Just google her.

Almost everybody here knows who she is, William. Don't assume ignorance.

Never heard of her; thanks for not assuming omnipotence.

If it's not heredity or shared environment, it could just be noise.

Like Sailer, I know of a pair of dissiimlar identical twins. My coworker is a systems administrator, his identical twin does graphic design.

I don't get the feeling that this is a disinterested debate.

It's not, because the data is out there in the form of test scores, grades, life experience, etc., across time and continents and points to a pretty clear conclusion: good genetics show up in all sorts of environments but the converse doesn't hold. That's why universities, colleges, businesses, employers, professions and couples using other people's baby batter all employ elaborate screening procedures instead of flinging the gates open and inviting all to partake of the enriching environment. The debate remains open because one side stays busy generating explanations for why their hypothesis can never actually be tested.

Seems like the "non-shared" environment is largely self-chosen. So how can we say that non-shared environment is a "major source" of environmental variance of personality, and so on? It seems like non-shared environment is more like "second-order heredity" rather than pure environment.

That is a problem, but it is hard to say that environment is chosen. I was in 1st and 2nd grade with one identical twin while his brother was in another. That wasn't chosen but peer groups at that age have a huge effect on many factors.

You might not get to pick the kids in your class, but you do have a say in who you play with at recess and sit with at lunch. Even holding the peers constant, you still decide how you interact with them. I would even go as far as to say that you have control over the extent to which you allow others to influence you.

I agree there are constraints, but I think there is also a lot of latitude within those constraints for individual choice.

We already know that peer groups and older siblings are by far the greatest influence on child development. See The Nurture Assumption. This piece adds nothing.

Clearly a Bryan Caplan directed post.

But wait, I'm confused: do these results show that there is no shared environment, or could the results be equally interpreted to mean that environment doesn't affect psychological characteristics?

Basically: learning happens. Small differences in experience become large differences in environment (and brain!) see for instance:

BTW IQ is unimportant in many of these discussions. Take a set of identical twins with very high genetic intelligence, physically good looking and charismatic. Put one in a USA home with the lowest 10% of income in a large USA city and one in a yuppy to 20% suburb, the one in city might learn just as much but different stuff and so do worse in the IQ test.

BTW they could rise to the top of their respective environment one to Harvard and one to run a garage or some other business.

Comments for this post are closed