[Edit: BlacK, White, and biracial classifications, in this analysis, include individuals who identify as Hispanic. See “NAEP, biracial trends” for an analysis which excludes Hispanics.]
Now and then I think I come up with some killer disproof of racial HBD and then latter discover that I botched my analysis. For example, last post, I deduced that a strong hereditarian hypothesis was untenable based on NAEP mixed race scores. I estimated mixed Black-White scores from the total (school) reported mixed scores based on what I thought was a reasonable assumption: that the Black-White mixed scores were not much lower than the total mixed scores. JL, however, pointed out that the new NAEP explorer — which is awesome by the way — allows for cross tabulations. It’s like SPSS for dummies. Using the cross tabs, one can estimate the mixed Black-White scores by comparing self-report with self-report or self-report with school-report (i.e., parent report). Estimates can be checked against one another. And then checked across grades and subject. Anyways, so I stand corrected. The NAEP mixed race scores are roughly consistent with a hereditarian model, at least in the case of the Black-White difference. (While Explorer doesn’t give sample sizes, one can backwards calculate them from the standard error and standard deviation. For the biracial White/Blacks in these samples, they range from ~90 (based on cross tab self and school reports) to ~800 (based on cross tab self and self report. As can be seen there is a broad consistency between mixed race scores calculated differently.)
[edited]
Within races, controlling for SES inevitably controls for IQ and genes. Take a look at this pic from a relatively recent paper. You can see that the IQs of biological offspring correlate positively with SES, but the IQs of adopted offspring do not. Findings like this and others indicate that a large portion of the correlation between IQ and SES, within populations, is a function of genes. In fact, it’s generally found, in behavioral genetic research, that SES and other environmental factors that vary between families explain little or no variance in IQ by adulthood. This is called the “gloomy prospects.” The upshot is that controlling for SES between races gives us little analytic advantage. If differences were genetic in origin, controlling for SES, would control for them. The assumption otherwise is called the sociologist fallacy.
Nonetheless, it would be interesting to see if the gap between Whites and mixed race kids increased with SES — like those between Whites and Blacks.
Sullivan made that point recently. I don’t find it convincing.
If you control for enough “SES” variables, you can narrow the gap quite a bit, especially at young ages. See: Yeung and Pfeiffer (2008) “The black–white test score gap and early home environment.”
The problem is that the more you control for, the more g you’re controlling for. And that controlling doesn’t have as large of an effect at older ages as it does at younger ages — presumably, because the narrow heritability of IQ increases.)