[Snargorf]

My old hypothesis is that in a reasonably free society, families reach their "destination" socioeconomic status after 3 generations.

Even if they immigrate as refugees on rusty boats, or lose everything in a holocaust or internment camps. Or, even if they win the lottery. After three generations, they hit their level, whether it's at the bottom or the top. And then they stay there.

It's IQ, an absence of stimulation-seeking behavior, a long mental time horizon, and a non-susceptibility to addictive chemicals or behaviors, and an absence of costly mental and physical diseases. It's genetic.

[sowhatquestion]

Correct me if I'm misreading, but I have a hard time seeing how the study supports your hypothesis. In the "Discussion" section, the authors state that their heritability estimate for educational attainment was 31%, and only 20% for SES. That still leaves the door open for significant social influence.

The authors summarize: "Our findings add weight to the view that genetic variation plays an important, but not exclusive, role in educational inequalities and social mobility, which is at variance with views, that still prevail in some quarters, that these are solely the product of social forces and environmental inequalities." In other words, they only take themselves to have refuted the strawman view that observed differences "are solely the product of social forces."

[cameldrv]

Worth reading: http://slatestarcodex.com/2016/03/16/non-shared-environment-...

He makes a lot of points here, but the most important IMO is that many measurements are noisy, and the correlation between X and Y+noise is less than the correlation between X and Y.

[gwern]

> In the "Discussion" section, the authors state that their heritability estimate for educational attainment was 31%, and only 20% for SES. That still leaves the door open for significant social influence.

Yes, that's true. However, GCTA has several limitations which mean that these numbers can be considered lower bounds on what the total genetic contribution: https://en.wikipedia.org/wiki/Genome-wide_complex_trait_anal...

[nradov]

Families are not genetically stable. There is very little inbreeding in Western societies, so a grandparent and grandchild may not be genetically close.

[Snargorf]

Families breed with other families similar to themselves.

[w1ntermute]

A very important part of "long mental time horizon" is picking a good spouse and raising your children well, traits often absent in those who shoot to wealth in one generation.