Monday, November 29, 2004
Nature v Nurture v Libertarian v Liberal
After reading this article on marginal revolution, I had planned on writing a post on this paper by Bruce Sacerdote. Sacerdote examined a large group of Korean children randomly assigned to adoptive parents in the US during the 70's and a group of children raised by their adoptive parents. Sacerdote then compared the two groups with respect to their parents on a number of factors - height, weight (obesity), proclivity to smoking, and so on.
This produced a number of results that you already knew like:
Your height is strongly influenced by your biological parents' height (nature)
Your probability of smoking is strongly influenced by whether or not you were raised in a household that smokes (nurture).
But the one result that has gotten everyone's attention (I'm using "everyone" very loosely) is the comparison of the children's income against their parents' income. The biological kids' incomes are strongly (and positively) correlated with their parents' incomes. However, the adopted kids show virtually no correlation with their adopted parents' incomes. This indicates that there's a strong correlation between genetics (or nature) and income.
If you read Gene Expression (and if you're not reading the site, you should be), you would know this is a subject they couldn't pass up (Gene Expression's unwritten motto seems to be "IQ is the best indicator of almost everything and IQ is largely determined by genetics" - a point I won't quibble with, they're just unusually upfront about it). So shortly after I read the post on marginal revolution, TangoMan already had a post up. He says about the same thing and directs his readers to the Nurture Assumption(which I haven't read). However he also offers this interesting observation.
Jane Galt (a libertarian site) and Kevin Drum (a liberal site, formerly CalPundit, now Washington Monthly's blog) have both linked to the same piece. We now have an interesting case study unfolding before our eyes as to how one's political leanings color one's reaction to a specific data point.
Reading the comments to the two posts have made for quite the amusing afternoon. Sometimes political labels DO have meaning.
Update
Sacerdote seems to have hidden the paper behind a pay for access wall. If you want a copy and I know who you are, email me, and we'll see what we can do...
The point of the plot was not the disparity of incomes between biological children and the adopted children, but the relative shape of the curves.
The biological childrens' incomes positively (and strongly) correlate with their parents' incomes and the plot of children's income versus parents' income slopes sharply up. This is an expected result and is traditionally attributed to these kids being a) exposed to the same type of peer groups b) afforded the same types of educational opportunities c) able to tap into their parents' network of connections.
The adopted childrens' incomes, however, exhibit virtually no correlation between their adoptive parents' incomes and their incomes hence their curve is practically flat. However, it is reasonable to assume that the adopted kids would also have access to a), b), and c). Even if the adopted kids have a massive number of disadvantages arrayed against them, then increasing parental wealth should be reflected in increasing adopted child wealth (adopted child wealth still should exhibit positive correlation with adopted parental wealth).
The key implication from this plot is there is clearly something that parents impart to their biological children in proportion to their income that parents cannot/do not impart to their adoptive children. There is one simple thing that biological parents give to their children that adoptive parents cannot give to their children - their genes. A genetic explanation neatly explains the relative slope of the two plots.
All other potential sources of explanation (racial, difference in treatment of biological and adoptive children) fail unless you suppose that those sources are applied in a manner that is inversely proportional to the nurture benefit imparted by increasing parental income.
A similar explanation is given in the follow-up over at marginalrevolution.
As a final note, commentors on other sites (and JP here) have questioned the data set sizes used to create the plot. I have taken the liberty of extracting the relevant table from the paper and have displayed it below. As you can see, reasonably sized samples were used for each data point.
|
This produced a number of results that you already knew like:
Your height is strongly influenced by your biological parents' height (nature)
Your probability of smoking is strongly influenced by whether or not you were raised in a household that smokes (nurture).
But the one result that has gotten everyone's attention (I'm using "everyone" very loosely) is the comparison of the children's income against their parents' income. The biological kids' incomes are strongly (and positively) correlated with their parents' incomes. However, the adopted kids show virtually no correlation with their adopted parents' incomes. This indicates that there's a strong correlation between genetics (or nature) and income.
If you read Gene Expression (and if you're not reading the site, you should be), you would know this is a subject they couldn't pass up (Gene Expression's unwritten motto seems to be "IQ is the best indicator of almost everything and IQ is largely determined by genetics" - a point I won't quibble with, they're just unusually upfront about it). So shortly after I read the post on marginal revolution, TangoMan already had a post up. He says about the same thing and directs his readers to the Nurture Assumption(which I haven't read). However he also offers this interesting observation.
Jane Galt (a libertarian site) and Kevin Drum (a liberal site, formerly CalPundit, now Washington Monthly's blog) have both linked to the same piece. We now have an interesting case study unfolding before our eyes as to how one's political leanings color one's reaction to a specific data point.
Reading the comments to the two posts have made for quite the amusing afternoon. Sometimes political labels DO have meaning.
Update
Sacerdote seems to have hidden the paper behind a pay for access wall. If you want a copy and I know who you are, email me, and we'll see what we can do...
The point of the plot was not the disparity of incomes between biological children and the adopted children, but the relative shape of the curves.
The biological childrens' incomes positively (and strongly) correlate with their parents' incomes and the plot of children's income versus parents' income slopes sharply up. This is an expected result and is traditionally attributed to these kids being a) exposed to the same type of peer groups b) afforded the same types of educational opportunities c) able to tap into their parents' network of connections.
The adopted childrens' incomes, however, exhibit virtually no correlation between their adoptive parents' incomes and their incomes hence their curve is practically flat. However, it is reasonable to assume that the adopted kids would also have access to a), b), and c). Even if the adopted kids have a massive number of disadvantages arrayed against them, then increasing parental wealth should be reflected in increasing adopted child wealth (adopted child wealth still should exhibit positive correlation with adopted parental wealth).
The key implication from this plot is there is clearly something that parents impart to their biological children in proportion to their income that parents cannot/do not impart to their adoptive children. There is one simple thing that biological parents give to their children that adoptive parents cannot give to their children - their genes. A genetic explanation neatly explains the relative slope of the two plots.
All other potential sources of explanation (racial, difference in treatment of biological and adoptive children) fail unless you suppose that those sources are applied in a manner that is inversely proportional to the nurture benefit imparted by increasing parental income.
A similar explanation is given in the follow-up over at marginalrevolution.
As a final note, commentors on other sites (and JP here) have questioned the data set sizes used to create the plot. I have taken the liberty of extracting the relevant table from the paper and have displayed it below. As you can see, reasonably sized samples were used for each data point.
|
