Many of our categories are human constructions which map upon patterns in nature which we perceive rather darkly. The joints about which nature turns are as they are, our own names and representations are a different thing altogether. This does not mean that our categories have no utility, but we should be careful of confusing empirical distributions, our own models of those distributions, and reality as it is stripped of human interpretative artifice.
I have argued extensively on this weblog that:
1) Generating a phylogeny of human populations and individuals within those populations is trivial. You don't need many markers, depending on the grain of your phylogeny (e.g., to differentiate West Africans vs. Northern Europeans you actually can use one marker!).
2) These phylogenies reflect evolutionary history, and the trait differences are not just superficial (i.e., "skin deep").
The former proposition I believe is well established. A group such as "black American" has a clear distribution of ancestries in a population genetic sense. The latter proposition is more controversial and subject to contention. My own assumption is that we will know the truth of the matter within the generation.
A black American
But that is the biological construction of race. Subject to fudge and fuzziness, but mapping upon a genuine reality. What about the social construction? Due to its flexibility this is a much more difficult issue to characterize in a succinct manner. Consider the cultural conditionals which render G. K. Butterfield "black" and Luis Guzman "Hispanic." Both individuals are products of an admixture between people of mixed African and European ancestry (and likely some Amerindian in Guzman's case). It turns out that the genes have segregated out such that Butterfield reflects more his European ancestry in traits. Guzman's phenotype is more mixed. The perception of these two individuals is weighted by two different strains in modern American racial ideology. First, that of hypodescent, where one drop of black blood means that an individual is black, without equivocation. Halle Berry appealed to this framework to argue why her daughter, who is less than 1/4 African in ancestry (Berry's African American father almost certainly had some European ancestry) was black. No matter that hypodescent's origins were to buttress white racial supremacy and purity. Today black Americans espouse for purposes of community solidarity (the black American community as we know it is a partly a product of hypodescent which forced mixed-race blacks into the African American community).
Not a black American
The second issue, which has crystallized in our time, but has roots back decades, is the peculiar position of "Hispanics/Latinos" in the American racial system. As A.D. Powell has observed Hispanics seem to be able to evade the one drop rule, unless their African features are extremely dominant (e.g., pre-skin whitening Sammy Sosa). I've looked at the genotypes of enough Latin Americans to assume that some level of African ancestry (e.g., ~5%) is present in the vast majority of those who are not the children of recent European immigrants or from indigenous communities. For example, Mexico's large slave population seems to have been totally absorbed, to the point where their past existence has been nearly forgotten. Mexicans of mestizo or white identity routinely have African ancestry, they just don't know it, nor is it part of their racial identity. And it isn't just Latinos. People of Middle Eastern ancestry, in particular Arabs, often have some African ancestry. But they are not classified as black (unlike Hispanics/Latinos they don't have their own ethnic category, but are put into the "white" box, irrespective of their race, from Afro-Arab to Syrian).
This broader coexistence of frameworks persists on the implicit level. We don't usually explicitly flesh out these details. Rather, we take these social constructions as givens. The major problem is when the problems and artificialities of these social constructions begin to bleed over into attempts to understand patterns of biological variation. Because of America's fixation on the black-white dichotomy rooted in skin color people routinely offer up the fact that the human phylogeny is not well correlated with pigmentation as a refutation of the concept of race. What biology is doing is refuting a peculiar social construction of race. It is not negating the reality of human population substructure. Sociology and culture anthropology are empires of imagination to a much greater extent than human biology.
I'm thinking of this because with the birth of my daughter I confronted the bleeding over of the social into the biological. For medical purposes her race had to be assessed. One side of her ancestry was not problematic; white European. But I had to argue for why her other half should not be listed as "Asian." For sociological purposes I have no great issue with the term Asian American which is inclusive of South and East Asians (I am not denying that this a recent political identity, I am saying that I do not personally find it objectionable and routinely enter my race as "Asian American" into public forms). But for biological purposes this is an incoherent and misleading classification. I know when my sister was born my parents put her race as "Asian," which even at the time I felt was totally without purpose as far as biological taxonomy went. At the end of it all my daughter had "South Asian" entered in by hand. Better that her information be discarded than aggregated into a data set in a misleading fashion.
Obviously disentangling the social and biological is not necessarily impossible. Rather, it takes a little care and explicitness, as it is so easy to move between the two domains so easily as to elide their differences. And to some extent they do inform each other. Personal genomics is adding a new twist, but the general problem is as old as human systematics. The only cure is care.
Image credit: Wikipedia