(Substantial prelude with some light technical bits, feel free to jump to [UPSHOT] or [BOLDFACE PUNCHLINE])
As is shown in the meta-analysis Andy references in his last post, more or less every measurable outcome anybody cares about in any study of human beings is more similar among identical twins than it is among fraternal twins, which in the classical model applied to twin study data means the trait has a non-zero heritability.
Perhaps the major motivation of the giant meta-analysis, however, is evaluation of the extent to which identical twin correlations are twice the fraternal correlation.
That finding can be interpreted as having an important implication for the amount of additivity in the genetic architecture of traits. Additivity is the extent to which the heritability of a trait reflects the sum of the influences of different genetic variants each considered on their own. The greater the extent of additivity, the stronger the rationale to do large-scale, extremely extensive work to try to identify causal genetic variants, even if their causal effects are very small.
One example of something that is an alternative to additivity are different kinds of gene-gene effects. However, for social scientists, a more obvious matter are effects of the so-called “shared environment,” which is the term behavioral geneticists use for all the exogenous environmental causes of a trait that cause siblings to be more similar. Under the assumptions of behavior genetics model, if the identical twin correlation is twice the fraternal twin correlation, this corresponds to a shared environment estimate of 0.
[UPSHOT] In other words, in addition to the meta-analysis underscoring the long-stated behavior genetics point that Everything Is Heritable, it also underscores their side point that Shared Environments Do Not Matter Much For Many Things.
About the latter issue far more could be said, but: I just wanted to mention that this finding about Shared Environments has a particularly prominent exception, which my former advisee Amelia Branigan demonstrated in a meta-analysis we co-authored awhile back. Educational attainment, so widely-studied by sociologists as both an independent and dependent variable, has a very strong shared environment component by conventional behavioral genetics methods. In fact, educational attainment, as far as I have been able to discern, is practically unique in terms of having such substantial heritability and shared environmental estimates.
So [BOLDFACE PUNCHLINE] the most studied non-demographic variable in sociology is also the biggest known exception to a major pattern that behavioral geneticists have observed time and again with other traits.
Why does educational attainment have such a high shared environment component, when cognitive test scores do not and when income measures do not? I have a paper with a current graduate student, Yu-Han Jao, that is working its way through the review process about that. We don’t answer the question, though, so much as explain why the answer is not obvious in the way one might think, and then try to evaluate the plausibility of some other contenders.