So stereotype threat exists but only explains a smallish part of the gap, with most of the rest due to genetics? ’kay.
Quantifying diversity is hard: genetic variation I don’t know (KHAN!), specific genes even less, ancestry data isn’t available, samples like “famous people” are skewed, etc. I mostly meant “Barack Obama: a definitely white and a definitely black parent, and he’s black in the US race system. That seems common”.
But here’s a way to test: pick people with a race system in common (typically the US one, and I could do an European replication). Ask them to describe their race (ideally open-ended, but given small samples probably a set list). Take pictures of them and ask a (blinded, racially sampled) jury to guess their race. Measure some objective and hopefully relevant criterion like melanin in skin, or some cleverly chosen gene, or ancestry if you have it handy. Have them do some kind of intelligence test. Possibly split into groups and test conditions like “stereotype threat”.
The mostly-genetics hypothesis predicts that the objective criterion will be the best predictor, and the jury estimation will be a better predictor than the self-report because it looks at phenotypical evidence of genome rather than irrelevant things like native language. The mostly-culture hypothesis predicts that the self-report will be the best predictor, and that the results will vary widely depending on local race systems.
Clever stupid “it’s all interaction” idea of the day: What about a genetic predisposition to social cues such as stereotype threats?
So stereotype threat exists but only explains a smallish part of the gap, with most of the rest due to genetics? ’kay.
Quantifying diversity is hard: genetic variation I don’t know (KHAN!), specific genes even less, ancestry data isn’t available, samples like “famous people” are skewed, etc. I mostly meant “Barack Obama: a definitely white and a definitely black parent, and he’s black in the US race system. That seems common”.
But here’s a way to test: pick people with a race system in common (typically the US one, and I could do an European replication). Ask them to describe their race (ideally open-ended, but given small samples probably a set list). Take pictures of them and ask a (blinded, racially sampled) jury to guess their race. Measure some objective and hopefully relevant criterion like melanin in skin, or some cleverly chosen gene, or ancestry if you have it handy. Have them do some kind of intelligence test. Possibly split into groups and test conditions like “stereotype threat”.
The mostly-genetics hypothesis predicts that the objective criterion will be the best predictor, and the jury estimation will be a better predictor than the self-report because it looks at phenotypical evidence of genome rather than irrelevant things like native language. The mostly-culture hypothesis predicts that the self-report will be the best predictor, and that the results will vary widely depending on local race systems.
Clever stupid “it’s all interaction” idea of the day: What about a genetic predisposition to social cues such as stereotype threats?
I don’t know if it “exists” or not. But clearly if it does exist it does not satisfactorily explain the gap.