A few days ago I posted a picture depicting the genetic ancestry of African-Americans modelled as a linear combination of European and Nigerian genetic clusters (reproduced below). Dienekes has some thoughtful comments on the same picture.
Meanwhile, Razib comments on a post from Ed Yong discussing the social flexibility of race. Ed cites a survey showing that people change their self-described race with surprising frequency, with about one in five respondents changing their description at least once over a nineteen year period (with some intriguing correlations with incarceration and employment status). Razib responds that the ability to "change" race is still contingent on biology; individuals with strong African ancestry are unable to readily pass themselves off as white.
It would be intriguing to quantitate this, by plotting genetic ancestry (as shown below) against the frequency of altering one's self-reported ancestry from year to year. Following Razib's argument, most African-American individuals clustering nearing the Nigerian end of the spectrum would be highly constrained as self-reported "black"; the further away from that end of the spectrum, the more freedom individuals would have to alter their racial identity to fit their social circumstances.
It would also be interesting to compare the effects of overall genomic ancestry with the impact of the comparatively small number of variants that determine racially salient features such as skin pigmentation; these would obviously be expected to have the largest impact on constraining self-described race.
Have similar studies already been done, perhaps using phenotypic markers rather than genetics? I'm completely ignorant about this area.
Goolge "construction of whiteness" and you will find piles of non-quantitative literature (but still, plenty of excellent thinking amid the morass) from scholars who have been thinking about this topic for the last decade.
Please also let's not forget that individuals who are not in one of these clusters (which themselves are artifacts of the data) need not be "racial hybrids." There are people living in their great great great great .. great grand daddy's village who fill in the gaps that appear to separate distinct patches on this chart.
Yuck... sorry, Greg, but "construction of whiteness" gives me one of the lowest signal-to-noise ratios of any Google search I've ever performed. Does anyone out there have quantitative data that would address this question? Extra points for links that don't contain the terms "discourse", "narrative", or "critical-interpretive methodology".
As for the clusters being "artifacts" - yes, the discreteness of the clusters is a consequence of the sampling strategy (only one to two populations sampled in each of the three outlying regions). Yes, human genetic diversity is largely continuous. But genetics still correlates extremely strongly with geographical ancestry. We might not be able to draw a clean line between "Europeans" and "South Asians" - but with high-resolution genotype data from enough long-term resident populations we will be able to pinpoint any individual's ancestral origins with surprising accuracy.
I know you're well aware of this, Greg, so I apologise for labouring the point - I just didn't want readers to be confused by the use of the term "artifact".
White. Skin. Privilege.
23andMe now lets you produce a similar graph for yourself using your SNP set and the SNP sets of your friends.
Related note: I have 32 people sharing my genome on 23andMe ---ALL of them have been initiated by the other member ---and other than a few of the 23andMe staff and some DNA network people, I don't know any of them ---nor have I ever communicated with them. Either I seemed particularly awesome in the like 10 total posts I ever made in the 23andMe forums, my profile picture is super hot and everybody wants to date me, or 23andMe is matching people in a mutually unsolicited MySpace-icebreaker-style.
But yah, bring it on. My 23andMe account name is "andrewyates". Share all you genomes with MEEEEeeee.