This, from @mendel_random and others. The authors show that correlations between genotype and phenotype (ie, heritability) can arise from at least 4 sources (there are more). Actual genetic causation is only one, and may not be the largest. https://advances.sciencemag.org/content/6/16/eaay0328

/1

They show that heritability can also be induced by popstrat, "dynastic" effects (rGE via parents) and assortative mating. All of these complications are especially likely for behavioral phenotypes. I think the consequences of this realization point in two directions. /2

On the one hand, genetic social scientists can produce endless studies pulling these strands apart. There is nothing wrong with this, but it won't lead anywhere definitive: in the absence of experimental manipulation in humans, the effort leads straight to the gloomy prospect. /3

On the other hand, realizing that hard genetic causation will never be separated from gloomy-prospect complexity puts hard limits on the strong conclusions of say, Robert Plomin or @charlesmurray about the social implications of the supposed "genetic revolution." /4

As I have said for years, genomics isn't going to turn behavioral science into biology; instead behavior is turning biology into social science. /end