When comparing greatly different sample sizes, (such as the dramatically different numbers of whites and blacks making >$100,000/year in the SAT sample), and particularly when dealing with the extremes of data sets, one generally needs to do further statistical analysis to correct for statistical artifacts due to the sample structure. Do you know if this has been done with the SAT data you reference? It doesn't sound like it.
It would be interesting to have more data on the parental literacy level and English proficiency and/or parental educational level, family status (two parent or one parent families, parent/grandparent/adoption/foster caregiver status of children), family size, and academic rigor of school attended for both black and white families, especially at the extremes of income in the SAT sample. "Income" clearly has its own impact on educational opportunity and family stress, but it would be interesting to examine whether there are really racial divides in the level of resilience and resistance to the negative educational effects of poverty as well as apparent resistance to the positive effects of wealth between families that are really very close in all other respects.
If we found, for example, that white families with very low incomes were much more likely to include a parent with proficient literacy skills, we wouldn't need to look much further to understand why their children had better educational outcomes. If we found that black children with family incomes higher than $100,000 had a much higher chance of having been adopted or fostered into the family, or to have been raised by relatives other than a parent, it might give us some insight into why the higher family income did not seem to provide the same benefit. We might, though, want to look at the reasons why those differences were there, and whether we as a society could eliminate or ameliorate some of the negative effects of the difference in circumstances and help identify and promote circumstances and behaviors that lead to great outcomes.
Certainly, IQ plays a role in achievement. IQ is highly heritable. This can lead to a certain fatalism regarding our ability to increase the level of literacy and educational achievement among lower income groups, and even to subtle and not-so-subtle racism and classism. There is, however, strong evidence that parental interaction style and language use in infancy and early childhood has a significant impact on IQ, particularly verbal IQ, and there is also evidence that parental interaction style and language use differs by both race and income level.
I think it is vitally important that we really examine this problem closely, because the economic drag and human impact of poverty and poor educational achievement on our society is staggering, and I believe that it is crucial that we recognize which factors are driving it, and which of those factors are within our power to change and which are not, so that we don't waste our resources on futile exercises, but we do take all the effective actions we can to deal with the issues that are under our control. Some of the apparent racial differences in achievement could be due to differences in family and educational circumstances that are not picked up by the data collection methods we are using. Others could be caused by real differences in IQ between racial and income groups that are nonetheless not due to genetic factors and that are open to change.
