« Why fewer Asian geniuses? | Gene Expression Front Page | Genes & politics »
December 01, 2004

The Adoption Controversy, Part Two

This is a follow-up to my post entitled The Nurture Assumption.

If you've been following the comments at Washington Monthly and at Jane Galt in reaction to the original post by Alex Tabarrok and his follow-up post on a paper entitled What Happens When We Randomly Assign Children to Families by Bruce Sacerdote (here is a free access version of the paper) you'll have noticed that many commentators are finding it difficult to accept Sacerdote's conclusion that parental income, while being correlated to the biological child's income, has virtually no influence on the income of the adopted children.

What I want to do in this post is to first clarify what I think is a calculated omission on Sacerdote's part and what is riling up the commentariat and secondly to point to other findings in Sacerdote's paper and see if they will be as controversial (warning - I'll write the second part with a straight face but the intent will be parody.)

It appears to me that Sacerdote's findings are being met with so much skepticism because he's hiding behind squid ink and the critics are finding it difficult to model the mechanism which would connect parental income to the income of the biological children but not to the adoptive children. It's obvious that the difference at work is parental genetic contribution to the biological children but I think the root of the skeptcism centers on the question of how can genes possibly control for income.

So, first, let's strip away some of the squid ink so we can assess the situation accurately. Sacerdote writes:

There is a large adoption literature outside of economics and it has focused mostly on estimating the heritability of IQ, as in Scarr and Weinberg [1978, 1981], and personality traits as in Loehlin, Horn, and Willerman [1985, 1987, 1994], and Plomin, Defries, and Fulker [1988, 1991, 1997]. I depart from this literature in two ways. First I focus on income, education and health outcomes rather than IQ and personality traits.

No, Sacerdote doesn't really depart from the literature by focusing on income, education and health. These factors are each related to g.

From Intelligence Predicts Health and Longevity, but Why?


O’Toole and Stankov used IQ at induction into the military, along with 56 other psychological, behavioral, health, and demographic variables, to predict noncombat deaths by age 40 among 2,309 Australian veterans. When all other variables were statistically controlled, each additional IQ point predicted a 1% decrease in risk of death. Also, IQ was the best predictor of the major cause of death, motor vehicle accidents. Vehicular death rates doubled and then tripled at successively lower IQ ranges. . .

IQ at age 11 had a significant association with survival to about age 76. On average, individuals who were at a 1-standard-deviation (15-point) disadvantage in IQ relative to other participants were only 79% as likely to live to age 76. . .

For each standard deviation increase in IQ, there was a 33% increased rate of quitting smoking. Adjusting for social class reduced this rate only mildly, to 25%. Thus, childhood IQ was not associated with starting smoking (mostly in the 1930s, when the public were not aware of health risks), but was associated with giving up smoking as health risks became evident. . .

Among diabetics, intelligence at time of diagnosis correlates significantly (.36) with diabetes knowledge measured 1 year later. Like hypertension and many other chronic illnesses, diabetes requires self-monitoring and frequent judgments to keep physiological processes within safe limits. In general, low functional health literacy is linked to more illnesses, greater severity of illnesses, worse self-rated health, far higher medical costs, and (prospectively) more frequent hospitalization.

From The Study of Human Intelligence: A Review at the Turn of the Millenium


Studies carried out in the US on the level of prediction of intelligence tests indicate that they are valuable instruments: "psychometric tests are the best predictors of success in school and in the world of work. And what’s more, they are no mean predictors of failure in everyday life, such as falling into poverty or dependence on the state (…). To say that other things are important, apart from intelligence, is not really a challenge until you say precisely what those other things are." According to the APA, standardised measures of intelligence correlate at levels of .50 with school performance, .55 with years of schooling, .54 with work performance, and –.19 with juvenile delinquency. No other psychological variable is capable of producing these correlations.


As far as the influence of SES is concerned, the APA report makes several points. It is more probable that the children of privileged families attain higher social status than those whose parents have low incomes or less education. Intelligence and family SES correlate at a level of .33. What is observed on comparing the occupational status (or income) of adult siblings raised in the same family and who, therefore, have the same family SES? In such cases, it is more probable that the brother or sister with the best intellectual performance in adolescence has the highest social status and the highest income in adulthood. It is also more probable that the brother or sister with best performance in IQ tests gets more out of education, so that s/he has better credentials for aspiring to a good job.

We see that g is correlated to health at 0.36, to income at 0.33, to work performance at 0.54, and years of schooling at 0.55. So to make the causative model clear let's substitute g for family income. Intelligence is highly heritable and this would explain a connection between the parent and biological child that isn't present for the parent and adoptive child. What critics are missing is that intelligence also correlates to education and income. So its not really the family income that is the causative agent but is in fact the intelligence of the parents.

Another confusing issue is the relatively flat line of the adoptee's income compared to the rising slope of the biological children's income. If the children are raised in the same family environment, then why the difference? Again we come back to the intelligence of the children as being the root cause.

Let's look at some data on IQ findings for a broad (though not comprehensive) cross section of jobs.

Jobs IQ.jpg

If we look at Household Income for 1980, we see that the top 5% of households earned $66,617 (current dollars) and the mean income for that year was $21,063 (current dollars.) If you compare the Household Income data for 1980 to the Sacerdote's graph you see that almost all of the parents are earning above the American Mean Income. This should be the first clue that we're not looking at a representative sample of parents.

Keeping in mind that there is a correlation between income earned and IQ, but also cognizant that the correlation is 0.33 and not 1.0, let's take a look at the distribution of the IQ range. An IQ of 125 will place you in the 95th percentile and the rarity of higher scores increases quite rapidly. Sacerdote doesn't provide us with direct IQ scores of the parents or chilren which would be extremely useful at this point but if we assume a moderate correlation between IQ and Income we can infer that a sizable majority of the parents have IQs that are higher than the American mean, some probably notably higher. I'll repeat, the parents are not a representative sample drawn from American society.

However, we have absolutely no reason to believe that the Korean adopted children are anything but a representative sample drawn from Korean society. Let's go back to the IQ distribution chart. Using a SD of 15, we see that 71% of the population falls within 2 SD of IQ 100. Various sources put the Korean mean IQ from 105-109.

So, it's quite likely that the Korean children were drawn from a representative sample of Korean birth parents and adopted by a non-representative sample of Americans who were earning above the American mean income, and likely had IQs well above the American mean of 100. We know that IQ is heritable (see Nature for the heritability of g at 0.8) and correlates to income. The biological children will also likely have IQs that are above the American mean and also above the Korean mean. There is a significant probability that their higher IQs will translate into increased earning potential and this will be correlated to the parental income. It should be noted that in 2003 the American Median Income was $43,318 (current dollars) and the adoptees, as noted in the study, are younger than the biological children, are earning right around the median income. Tabarrok notes the age difference between the adoptees and biological children:

Once you control for this and a few similar factors the mean income difference goes away (think of shifting the adoptee line up). What remains, and this is the key point, is that the biological line is upward sloping and the adoptee line is flat.

In other words, the adoptees, if they were truly drawn from a representative population in Korea are already earning a mean income. They are not subpar in this regard.

Now for those who still find this controversial perhaps you can work up the same skepticism for the Sacerdote's findings on height and Body Mass Index:

For height and obesity, there is strong transmission from parents to their biological children and almost no transmission of these outcomes from parents to adoptees. For example, the transmission coefficient on body mass index is .02 for adoptees and .23 for non-adoptees.

We also know that child-parent heights are correlated at 0.47 (note how that compares to IQ as a predictor of work performance 0.54) (See here for further information.)

Now why is it that adopted children raised in the same household as the biological children have absolutely no correlation to the height or Body Mass Index of the parents? Afterall, they are witness to family eating rituals, ingest the same nutrients as the biological children, subjected to the same food phobias and cravings, as well as the same exercise regimes and lifestyle choices and yet if we graphed these results we'd see absolutely no relationship despite these overwhelming environmental influences.

Intelligence is more strongly heritable than height or BMI. If you accept Socerdate's findings on height and BMI, what's the problem with intelligence having influence on family income and intelligence being heritable and enabling the biological children to earn incomes in relation to how well their parents earned? We have no reason to believe that the adopted children are uniformly drawn from the upper regions of the Korean IQ distribution so their performance is most likely on par for their IQ.

Posted by TangoMan at 03:25 AM