Thursday, July 27, 2006

OK so big news from Josh Marshall made me interupt my vacation from blogging
(also I am upset that no one and I mean not one single person complained).

Also I raised the issue of the heritability of intelligence. This is a very very delicate topic so I will plunge right in.

Kevin Drum has an interesting post presenting data which seems to contradict his analysis.

He notes that twin studies suggest that, if variation in IQ can be decomposed into hereditary and environmental components (a big if) then the hereditary component is larger than the environmental componenent. In fact, they suggest that it is slightly larger -- on the order of 60% of the variance. He argues that this understates the amount of IQ variation due to environmental factors since twin studies look only at adopted people and most adopters are middle to upper class. This is a good point (it is also worth noting that some of the most convincing studies which weren't completely fraudulent were conducted in Scandanavia which is not the place to look for effects of inequality). Thus one would guess that, if a decomposition makes sense, the environmental component is larger than the hereditary component.

Then he notes a study which does not use twins and but looks at IQ based on the very rough socioeconomic class of biological and adoptive parents. The study finds that the IQ's of people with poor biological parents and non poor adoptive parents are very slightly lower than the IQ's of people with non poor biological parents and non poor adoptive parents. Drum seems to think this shows environment is more important than suggested by twin studies. It doesn't. To the extent that it suggests anything it suggests that variance due to congenital factors is slightly greater than variance due to post adoption environmental factors. This is in line with twin studies.

So what about Drum's very convincing argument which suggests that twin studies underestimate the relative importance of environment compared to genetics ? I think the explanation is based on the difference between "congenital" and "genetic" that is "present at birth" and "heritable indefinitely at least for millions of years". Congenital factors which are not genetic include maternal malnutrition, maternal smoking, maternal alcohol consumption and maternal crack consumption and maternal expose to lead. There are many factors which differ between the biological offspring of the poor and non poor which have nothing to do with genes. Each of the factors I mention is strongly associated with both IQ and mothers' income.

As to "if variation in IQ can be decomposed into hereditary and environmental components" this is only possible if IQ is the sum of a part due to genes and a part due to the environment, that is, if the effects are additively seperable. Clearly this is not true if any environmental factor is a function of a genetic factor. Consider IQ in Southern states of the USA for people who attended school before (at the earliest) Brown vs Board of Education (to be reasonable in all US States at any time up to the present) or in Apartheid South Africa. Skin color is definitely highly heritable (has to do with exposure to the sun too). It affects the budget of the school one attented if one lived in those places at those times. That budget affects measured IQ (look at IQ by US state if you doubt this). Thus IQ was highly heritable then and there because race as defined there and then is 100% heritable. I know I didn't notice this obvious point on my own. I think I may have gotten it from Jonathan Chait but then again maybe not.

update: See also this which I didn't notice the first time around and which only appears because the editors is whomping on Andrew Sullivan. Also just this second I notice that Mark Klieman makes the point about congenital and genetic somewhat more concisely than I did (and with no big words).

"Kevin Drum is right to say that studies of separately adopted twins have long been considered the gold standard in research on the heritability of IQ. But I've never understood why. Each twin in a pair spends nine crucial months in precisely the same environment, and an environment whose quality varies strongly with social class."

To be fair to twin studiers who didn't just make the data up (that is who are not Sir Cyril Burt) they compare fraternal and identical twins. So they look at pairs all of whom shared a womb and all of whom were raised by different parents some of whom had all the same genes and some of whom have half the same genes and half different genes. Differences in experience in the womb do not appear either as genetic or as environmental differences. Thus the ratio is post natal environmental vs genetic out of the sum of the two. The variance due to prenatal environmental factors is not estimated in twin studies. Thus my idea as to why the results from the new studies aer similar to results from twin studies even though Drum's criticism of twin studies is convincing. There is a similar problem in the new study which is that it lumps 9 months of crucial environment with genetics (not that there is anything the authors cold have done about it or any hint that they didn't understand and discuss the issue).

Anonymous has left a new comment on your post "7/27/2006 11:56:00 PM":

This assumes that intelligence can be measured. We can all site examples of people who are intelligent about some things and stupid at others. I understand the scientific need to study and measure things but when it comes to the brain, they have a long way to go. When building on previous knowledge, it is important to have a solid foundation to stand on and not just pretend that you do. By now it should be clear to everyone that the brain especially at a young age has enormous potential for learning.

yes Anonymous that is certainly true. Why at this site alone I can cite two examples of people who have interesting things to say but can't spell to save our lives. More generally scores on IQ tests are useful in predicting life outcomes. The tests measure something that matters. This does not mean that it corresponds to the ordinary meaning of the word "intelligence" but there is solid evidence that IQ test scores are correlated with something interesting and important.


?
Tilted Edge has left a new comment on your post "7/27/2006 11:56:00 PM":

This assumes that intelligence can be measured. We can all site examples of people who are intelligent about some things and stupid at others. I understand the scientific need to study and measure things but when it comes to the brain, they have a long way to go. When building on previous knowledge, it is important to have a solid foundation to stand on and not just pretend that you do. By now it should be clear to everyone that the brain especially at a young age has enormous potential for learning.



Also some very smart people have trouble dealing with blogger sometimes.


Anonymous has left a new comment on your post "7/27/2006 11:56:00 PM":

If you are ever have a psychological evaluation, one of the first questions you will be asked is "Did you Graduate from college?" In my opinion this should have nothing to do with your mental state but statistically they find that college graduates have better mental health so they ask you this question. This is bullshit.


?
Tilted Edge has left a new comment on your post "7/27/2006 11:56:00 PM":

Yes and legibility in handwriting skills surely are not a measure of intelligence.

No comments: