“Socioeconomic status modifies heritability of IQ in young children”.
To make a long story short, this analysis of a large cohort of children assigned a ‘socioeconomic status indicator’ to each family they were following from 0 to 100 based on a large number of factors. They found that the heritability of IQ was a VERY strong positive function of socioenomic status. At the bottom, they think less than 5% of IQ variation is moderated by genetics. At the top of the scale, over 80%.
Obvious interpretation: low socioeconomic status masks genetic predisposition. Alternative restatement: high status environments allow previously cryptic variation to show itself. Low status populations are too genetically diverse for there to be a common factor that doesn’t vary between any of them.
This is, of course, exactly the kind of result that you would expect to get given the way that heritability is defined. When you make environment more uniform in terms of quality, you drive up the heritability, and vice versa.
The strength of the effect is still interesting, though.
Doesn’t the Turkheimer et al. result suggest that equalizing environments can drive heritability up or down, depending on how one does it? It’s as if the norms of reaction converge with decreasing SES, whereas the usual heritability analysis implicitly assumes parallel norms of reaction.
Presumably you’re asking about variation as a function of SES...? If so, one can eyeball an answer from the top row of figure 2. At the bottom of the SES scale, there’re a bit under 500 units (IQ points squared?) of full scale IQ variance (none from additive genetic effects, a bit under 300 units from common environment, and a bit under 200 units from nonshared environment). At the top of the SES scale, there’re about 300 units of variance (essentially all coming from additive genetic variance). Note that these numbers all come with wide error bars.
Thought I would repeat something I recently posted buried deep in a digressionary comment thread of an old post:
http://pss.sagepub.com/content/14/6/623.short
http://dl.dropboxusercontent.com/u/67168735/heritability%20of%20iq.pdf
“Socioeconomic status modifies heritability of IQ in young children”.
To make a long story short, this analysis of a large cohort of children assigned a ‘socioeconomic status indicator’ to each family they were following from 0 to 100 based on a large number of factors. They found that the heritability of IQ was a VERY strong positive function of socioenomic status. At the bottom, they think less than 5% of IQ variation is moderated by genetics. At the top of the scale, over 80%.
Obvious interpretation: low socioeconomic status masks genetic predisposition. Alternative restatement: high status environments allow previously cryptic variation to show itself. Low status populations are too genetically diverse for there to be a common factor that doesn’t vary between any of them.
This is, of course, exactly the kind of result that you would expect to get given the way that heritability is defined. When you make environment more uniform in terms of quality, you drive up the heritability, and vice versa.
The strength of the effect is still interesting, though.
Doesn’t the Turkheimer et al. result suggest that equalizing environments can drive heritability up or down, depending on how one does it? It’s as if the norms of reaction converge with decreasing SES, whereas the usual heritability analysis implicitly assumes parallel norms of reaction.
Is the total amount of variation the same in different populations? How does the magnitude of the variation contributed by genetics compare?
(That is to say, one percentage may be very different from another percentage.)
Presumably you’re asking about variation as a function of SES...? If so, one can eyeball an answer from the top row of figure 2. At the bottom of the SES scale, there’re a bit under 500 units (IQ points squared?) of full scale IQ variance (none from additive genetic effects, a bit under 300 units from common environment, and a bit under 200 units from nonshared environment). At the top of the SES scale, there’re about 300 units of variance (essentially all coming from additive genetic variance). Note that these numbers all come with wide error bars.
Yep, that’s what I was asking about. Thanks!
I put a HBD chick into play.