In Selfish Reasons to Have More Kids, I showed that nurture effects are small within the First World.  But I also freely conceded that the nurture effects of growing up outside the First World are probably large:

The most important weakness of behavioral genetics, though, is simply that research focuses on middle-class families in First World countries. The results might not generalize. Twin and adoption studies almost never look at people in Third World countries. So you shouldn’t conclude that Haitian orphans would turn out the same way if raised in Sweden.

[…]

Twin and adoption research only show that families have little long-run effect inside the First World. Bringing kids to the First World often saves their lives. Over 13 percent of the children in Malawi–the African nation than initially denied Madonna’s petition to adopt a four-year-old orphan–don’t survive their first five years. And survival is only the beginning. Life in the First World spares children from hunger, disease, and harsh labor, and opens vast opportunities that most of us take for granted. Merely moving an adult Nigerian to the United States multiplies his wage about fifteen times. Imagine the benefit of giving a Nigerian child an American childhood and an American education.

During the last month, I’ve delved much more deeply into this subject.  As you might expect, there is a sizable literature on the effects of international adoption.  The evidence on physical benefits is strong and clear-cut.  Toddlers adopted from the Third World are tragically below normal on height, weight, and head circumference on arrival.  (d is effect size in standard deviations).  Within eight years, however, these adoptees close about 75% of height and weight deficits and about 35% of the head circumference deficit.

adoption.jpgBut what about intellectual benefits?  Average IQs in Third World countries are quite low – mid-to-high 80s for Latin America, mid-80s for the Middle East, low-80s for South Asia, 70s or even lower for sub-Saharan Africa.

Before we turn to the evidence, however, there’s a major complication.  In the First World, children adopted by smarter parents get somewhat higher IQ scores.  Unfortunately, these benefits fade-out by adulthood.  Me again:

Twin and adoption research on young children’s intelligence always finds nurture effects. The younger the child, the more parents matter. A team of prominent behavioral geneticists looked at major adoption studies of IQ. They found moderate nurture effects for children, versus none for adults. Suppose an adoptee grows up in a family with a biological child at the 80th percentile of IQ. During his childhood, we should expect the adoptee to have a higher IQ than 58 percent of his peers. Nurture effects were largest for the youngest kids under observation, four-to six-year-olds. An average child of this age raised in a high-IQ home will typically test higher than 63 percent of his peers. Not bad–but it doesn’t last.

The upshot is that measuring the IQs of Third World adoptees when they’re children isn’t very informative.  Even if they show large gains, the gains could easily be fleeting.  Instead, we should wait and measure those kids’ IQs when they’re adults.

I had to read a pile of papers, but in the end, I found what I was looking for.  Dalen et al.’s “Educational Attainment and Cognitive Competence in Adopted Men” (Children and Youth Services Review, 2008) looks at Swedish conscripts’ cognitive test results for (a) 342,526 non-adopted Swedes, 780 adoptees from South Korea, and 1558 adoptees from other non-Western countries.  Subjects were roughly 18 years old when they took the test.  At my request, Bo Vinnerljung, one of the authors, shared the nationality breakdown for the “other non-Western” countries: India (21%), Thailand (19%), Chile (13%), Sri Lanka (9%), Colombia (9%), Ethiopia (8%), Ecuador (7%), with the remainder from “a wide mix of small groups, e.g., Poland, Peru, Bolivia, a few from other sub-Saharan Africa countries, and a small group from Mid-East countries including Iran.”

Suppose you plug in the most commonly-used estimates of IQ by country, assign each of Vinnerjung’s “small groups” equal shares of the remainder, and assume international adoptees are average for their home country.  Then if they’d stayed in their birth countries, the other non-Western adoptees would have a mean IQ of only 84.

So how smart were these other non-Western adoptees at maturity (approximately 18 years old)?  Let’s normalize the scores so overall Swedish IQ is 99 (a typical estimate), with a standard deviation of 15.*  Then other non-Western adoptees’ average score was 88.  Four IQs points may sound modest, but it’s one of the biggest environmental effects on adult IQ in the literature.

The real magic happens, though, when we look at the breakdown by age of adoption, provided in this companion paper by Odenstad et al.’s “Does Age at Adoption and Geographic Origin Matter?” (Psychological Medicine, 2008).

Other Non-Western IQ by Age of Adoption

Age at Adoption

IQ

0-6 months

90

7-12 months

88

13-18 months

89

19-24 months

89

2-3 years

87

4-5 years

85

7-9 years

76

Since adoptees’ biological families are generally underachievers even by Third World standards, the true IQ benefit of adoption at birth is probably even greater than it looks.  Adoptees raised by their biological families could easily have had an average IQ of only 80, implying that early adoption eliminated over half their cognitive deficit.*

At this point, you may be asking, “Wait, what about the Korean adoptees?”  They did slightly better than regular Swedes, but markedly worse than regular South Koreans (average national IQ: 106).  Their results by age of adoption show no clear pattern:

Korean IQ by Age of Adoption

Age at Adoption

IQ

0-6 months

100

7-12 months

98

13-18 months

106

19-24 months

100

2-3 years

103

4-5 years

97

7-9 years

99

If you’re using international adoption data to test the view that international IQ differences are 100% environmental, the contrasting performance of Korean and non-Korean adoptees is bound to be disturbing.  Sure, you can blame all remaining gaps on pre-natal environment.  But if equalizing a vast array of post-natal experiences leaves a big gap, what makes you so sure that equalizing pre-natal experiences would make the gap vanish?

If you’re using international adoption data to test the view that international IQ differences are 100% genetic, however, the performance of the non-Korean adoptees is powerful testimony to the wonder of First World upbringing.  Suppose the earliest adoptees are average for their home country.  Then growing up in Sweden raises their adult IQs by .4 SDs – 40% of the huge IQ gap between Sweden and adoptees’ countries of origin.  And since even the earliest adoptees are likely below-average for their home country, the true gain is probably larger still.  International adoption doesn’t make international IQ gaps vanish, but it plausibly cuts them in half.  And remember – unlike classic childhood interventions like Head Start, the gains last into adulthood instead of fading away.  What other viable, lasting treatment for low IQ is even remotely as effective?

* It is somewhat tempting to assume that if all of these kids had stayed with their biological families, the whole sample would look like the 7-9 year-olds – or worse.  But we should resist this temptation.  Many of the 7-9 year-old non-Koreans probably grew up in hellish orphanages, which seem noticeably worse for IQ than the typical Third World home.  Furthermore, late adoptees plausibly had extra developmental problems that partially caused their delayed adoption.