> As I recall, the IQ heritability research was done with twin studies.
No, it wasn't. Such a set up is impossible because of lack of samples and sampling bias.
All that "IQ is X% heritable" means is that they ran a linear regression on a large data set and saw that the variance error in the regression reduced by X% when they plugged in parental IQ as a covariate in the regression.
This is a correlational study as causal studies are impossible. Correlational studies are inherently spurious and ignore lots of confounding factors. Nevertheless, we can say that what we routinely observe in real life - smart parents having smart kids, has been grounded in real data.
For getting effect of race on IQ you plug in race as a covariate. Typically, most of the race based IQ research remove parental IQ as a covariate. All of these regression based studies are dubious and there is no real way to correct it. A large data set of twins who were separated at birth could help, but there are caveats there as well.
As an aside, Jordan Peterson seems to love these race IQ regression models and likes to call them hard undeniable science. He immediately switched sides when talking about climate models and how these models can't be trusted, because the simple choice of covariate included creates absolutely unreliable and biased models.
No, it wasn't. Such a set up is impossible because of lack of samples and sampling bias.
All that "IQ is X% heritable" means is that they ran a linear regression on a large data set and saw that the variance error in the regression reduced by X% when they plugged in parental IQ as a covariate in the regression.
This is a correlational study as causal studies are impossible. Correlational studies are inherently spurious and ignore lots of confounding factors. Nevertheless, we can say that what we routinely observe in real life - smart parents having smart kids, has been grounded in real data.
For getting effect of race on IQ you plug in race as a covariate. Typically, most of the race based IQ research remove parental IQ as a covariate. All of these regression based studies are dubious and there is no real way to correct it. A large data set of twins who were separated at birth could help, but there are caveats there as well.
As an aside, Jordan Peterson seems to love these race IQ regression models and likes to call them hard undeniable science. He immediately switched sides when talking about climate models and how these models can't be trusted, because the simple choice of covariate included creates absolutely unreliable and biased models.