The Blacks, the Whites, and James Watson

By now you've likely heard about DNA pioneer James Watson's remarks. If not, here is a quick recap from the horses mouth, via The Independent:

"[Watson] says that he is "inherently gloomy about the prospect of Africa" because "all our social policies are based on the fact that their intelligence is the same as ours - whereas all the testing says not really."

Well, let's see what the numbers actually say.

I reviewed the recent research on the measured gaps (test scores, wages, skill attainment) between blacks and whites in the United States. The general thrust of this work (see here, here, and here) is that predictive evidence for future gaps exists at the time when schooling first starts, so if there is a non-genetic reason for differences in intelligence, one would think they must occur very early in a child's life.

But how to test for this? As with any empirical research, it means finding a new dataset.

The U.S. Department of Education began collecting data on some 10,000 children born in 2001. One piece of this was a test of cognitive abilities applied to children between 9-12 months old and then again at the age of two. The children and their parents will continue to be surveyed through the first grade.

Roland G. Fryer of Harvard and Steve Levitt of the University of Chicago used the data accumulated thus far to see if there were any measurable differences at these very early ages.

Some notable points from the DOE's survey, via Fryer and Levitt:
- white infants scored 0.013 standard deviations better than average on the mental measure at 9 months and 0.176 standard deviations better at age two
- Roughly 90 percent of white and Asian infants are living in households with two biological parents compared to only 41 percent among Blacks.
- Asian children have slightly fewer siblings than average while black children have slightly more.
- white and Asian mothers tend to be older.
- white parents fare better than the other groups on the interviewer evaluation of "parent as teacher" effectiveness.
- Extremely low birth weight and premature birth are most common among blacks and least frequent for whites and Asians.

Two graphics below from their paper show visually how differences in measured abilities start to show up as of the second round of testing.

The first graphic shows mental test scores at 9 months which reveal very little variation among different ethnic groups. The second chart however shows that white test scores have shifted to the right (i.e. higher test scores).

From Fryer and Levitt's working paper:

"Analyzing these data, we find extremely small racial differences in mental functioning of children aged eight to twelve months. With no controls at all, the mean White infant outscores the mean Black infant by .055 standard deviation units - only a tiny fraction of the one standard deviation racial gap typically observed at older ages. The raw scores for Blacks are indistinguishable from Hispanics and Asians, who also slightly underperform Whites."

After controlling for things like gender, socioeconomic status, home environment and prenatal circumstances, the researchers find no statistically significant difference between the test scores of whites and other ethnicities.

Fryer and Levitt, however, say that while it looks like there is no cognitive differences between children of different ethnicities very early in life, they can't rule out that genetic differences manifest themselves later on.

They then try to come up with a model which best fits the observations in the data (that test scores are nearly identical close to birth but start to diverge as children grow up). Among three models they test out the one that fits best is the one in which innate mental ability remains the same over a person's lifetime while environmental influences grow more important with age. (The other models they test are 1) innate abilities and environmental impact are constant through time 2) innate abilities grow over time while environment stays the same.)

So while Fryer and Levitt do the scientifically proper thing by not saying that their research definitively rules out innate differences between various ethnicities, I'm very much inclined to.

(Hey, a whole post on 'blacks' and 'whites' without using the word "race".)