Hum Genet (2009) 126:355–362

DOI 10.1007/s00439-009-0674-1

REVIEW ARTICLE

Looking for race in all the wrong places: analyzing the lack of productivity in the ongoing debate about race and genetics

Morris W. Foster

Received: 2 February 2009 / Accepted: 17 April 2009 / Published online: 25 April 2009© Springer-Verlag 2009

Abstract The ongoing debate about the relationshipbetween race and genetics is more than a century old andhas yet to be resolved. Recent emphasis on population-based patterns in human genetic variation and the implica-tions of those for disease susceptibility and drug responsehave revitalized that long-standing debate. Both sides in thedebate use the same rhetorical device of treating geo-graphic, ancestral, population-speciWc, and other categoriesas surrogates for race, but otherwise share no evidentiarystandards, analytic frameworks, or scientiWc goals thatmight resolve the debate and result in some productive out-come. Setting a common goal of weighing the scientiWcbeneWts of using racial and other social heuristics with test-able estimates of the potential social harms of racializationcan reduce both the unreXexive use of race and other socialidentities in biological analyses as well as the unreXexiveuse of racialization in social critiques of genetics. Treatingsocial identities used in genetic studies as objects for inves-tigation rather than artifacts of participant self-report orresearcher attribution also will reduce the extent to whichgenetic studies that report social identities imply that mem-bership in social categories can be deWned or predictedusing genetic features.

Introduction

Scholarly debate about the relationship between geneticsand race (as well as other forms of social identity that implyancestry) has a long history (Boas 1912; International

Congress of Eugenics 1923). The current revitalization ofthat debate, triggered by an emphasis on human geneticvariation emerging from the Human Genome Project(Chakravarti 1999), provides additional evidence of thecomplexity and seeming intractability of the scientiWc ques-tions at its heart, as well as the many social, political,moral, public health, and other issues that have becomeentwined with it.

For the most part, the historical discourse has beenframed by the assertion of rhetorical absolutes: race hasbiological meaning and so must be attended to in geneticand biomedical contexts (Burchard et al. 2003; Risch 2006)or race has no biological meaning whatsoever and shouldbe erased from genetic and biomedical contexts (Lewontin1974; Goodman 2000). That neither of these competinghypotheses has achieved a consensus over the course of thelast century—at the same time that we exponentially haveincreased our knowledge of human genetics and humanvariation—suggests that the way in which the debate hasbeen framed insures that it will continue to be reproducedrather than resolved.

Indeed, while the debate has led to a multitude of publi-cations about race and genetics (including the present one),has contributed to the emergence of at least one academicdiscipline (anthropology), and has generated a great varietyof popularizations of both points of view (such as the cur-rent GenoGraphic Project and the American Anthropologi-cal Association’s Race: Are We So DiVerent?), it has notsigniWcantly altered research or clinical practices. At thebeginning of the 21st century, self-reported racial and othersocial identities continue to be used in recruiting partici-pants for and reporting the Wndings of genetic studies aswell as in providing clinical guidance for the diagnosis andtreatment of disease in much the same uncritical mannerthat those identities were used at the beginning of the 20th

M. W. Foster (&)Department of Anthropology, University of Oklahoma, 455 W. Lindsey, Rm. 505C, Norman OK 73019, USAe-mail: morris.w.foster-1@ou.edu

123

356 Hum Genet (2009) 126:355–362

century. That lack of productivity is perhaps less due to theintransigence of the scientiWc and medical establishmentsand more because the terms of the debate itself have beenframed by both sides in a way that uses “race” to distancebiological perspectives from social perspectives and viceversa, even as biological and social methods and data havebecome more sophisticated and detailed. We continue toengage in a black or white, two-dimensional argumentabout a multi-hued, multi-dimensional question.

An analysis of competing claims in the current iterationof the debate shows that the two polarized perspectives arebased on independent, parallel logics with quite diVerentevidentiary standards. At the same time, both views engagein the conXation of racial categorization and population-speciWc sampling to construct their arguments. Understandingthose diVering premises along with the mutual conXation ofrace and population sampling may help us Wnd a more pro-ductive (rather than self-reproductive) way of framingquestions about the relationship between genetics andsocial identity, which is the general domain within whichthe issue of race presents a special problem.

A primary reason for the unproductive nature of thedebate so far has been the tendency for the special case tooverwhelm the general domain. While social critiqueslargely have driven the polarization of the current iterationof the debate by making the special case for the racializa-tion of genetic research, a smaller number of provocativeclaims for the biological signiWcance of race (Risch et al.2002) also have contributed, as has the practice of uncriti-cally using social identities in relation to genetic data andanalyses on the part of many geneticists who otherwisehave not engaged in the debate. Consequently, althoughmost geneticists recognize the scientiWc complexity sur-rounding the relationship between genetic data and analy-ses and racial and other social identities, the debate itselffor the most part does not reXect that scientiWc complexity,for example over-generalizing the pattern of human geneticvariation, particularly within continents and subcontinents.

While the debate about race and genetics is not central toscientiWc advances in population genetics, it is arguable thatits continuation has perpetuated the unreXexive practiceswith respect to social identities on the parts of many geneticsresearchers, who have chosen not to engage with the unpro-ductive controversies that the debate has generated.

Biological claims and practices

Arguments for the biological relevance of racial and othersocial categories that imply ancestry make use primarily ofpopulation-based epidemiologic data and quantitativeanalysis of those data to show that race and other socialcategories account for signiWcant diVerences in biological

features such as frequencies of genetic variants (Burchardet al. 2005). In eVect, the argument is that racial and othersocial identities are predictive of evolutionary or demo-graphic population histories. An additional argument thenis required to demonstrate the claim that those histories arerelevant to questions of disease susceptibility and drugresponse, again using quantitative analysis to demonstrateassociations with disease incidence and drug response(Choudhry et al. 2006).

Current claims for the biological validity of racial andethnic categories do not include every single person whoasserts those social identities or on whom those identitiesare imposed. Rather, they apply to aggregate analyses ofpopulations with those social labels. In a globalized world,human populations are analytically signiWcant because ofthe diVering genetic histories they imply, evidenced bydiVerent frequencies of the same common variants, and thepresence of rare variants in some populations and not oth-ers. Consequently, population histories can be important indesigning genetic studies so that samples of cases and con-trols in association studies have similar background varia-tion to reduce the number of false-positive Wndings(Marchini et al. 2004).

Race and ethnicity traditionally have been used as surro-gates for homogeneity in that background variation tostratify populations for genetic analysis. The analytic problemof population stratiWcation has been a focus of recentclaims for the biological signiWcance of race and ethnicity,with proponents attempting to show that participant self-identiWcation accurately accounts for higher frequencies ofgenetic variants associated with particular population histo-ries (Tang et al. 2005). While other analyses have foundthat self-reported race and ethnicity are not predictive ofpopulation stratiWcation (Campbell et al. 2005), lower-cost,higher-throughput whole genome assays increasingly haveproven to be a more accurate and cost-eVective way toascertain stratiWcation among study participants, includingstratiWcation due to Wner-grained population histories thatare much less likely to be reported by participants (Tianet al. 2008). The use of genotyping to ascertain populationstratiWcation removes one of the primary analytic argu-ments for the utility of race and other social identities ingenetic research.

Markers that are informative for purposes of populationstratiWcation in the context of particular samples for spe-ciWc studies, though, also can be treated as informative formore absolute genetic deWnitions of ancestry. Thus, recentstudies have contrasted “self-reported ancestry” with“genetic ancestry,” where the latter is deWned by polymor-phisms that are more frequent in some geographically speci-Wed populations than others (Yang et al. 2008). However,rather than generating genetically speciWc categories suchas haplogroups (which then could be characterized as being

123

Hum Genet (2009) 126:355–362 357

in greater or lesser frequency in populations labeled withparticular social identities), some researchers use the samesocial labels and categories for “genetic ancestry” as thoseused for participant self-report of ancestry (Halder et al.2008; Klimentidis et al. 2009). Thus, both participant self-report and genetic ancestry are treated as two diVerent waysof deWning “African” or “European,” equating genetic deW-nitions of identity with social deWnitions.

Although most researchers who use social identities inreports of genetic Wndings do not endorse explicit claimsthat racial and other social categories are biologically deW-ned, the unreXective use of those categories in reportinggenetic Wndings also links the former with the latter in waysthat imply a direct relationship. For the most part, scientiststend to take racial and other social identities as given inreporting study Wndings, without further investigating howthose categories have been shaped by particular historicaland social conditions, why research participants use them tosituate themselves in the current social landscape, whetherreporting those identities contributes to the scientiWc valueof the genetic Wndings, or how Wndings of genetic diVer-ences associated with those identities could exacerbate bothmedical and popular misconceptions about genetic contribu-tors to health disparities among ethnic groups (Sankar et al.2007). Arguably, reducing the considerable detail of geneticdata and analyses to existing social labels for human popula-tions, whether by direct equation with “genetic ancestry” orby the unreXexive use of social identities, signiWcantlydevalues the scientiWc power of population-based geneticWndings. Such reductionism also devalues the power ofincreasingly detailed social data and analyses to which thosegenetic Wndings may be related through sophisticated mod-els of gene–environment interaction.

The social critique

While claims for the biological meaning of race based onepidemiologic data and quantitative analyses have beencontested using the same evidence and analytic methods byother epidemiologists and geneticists (Cooper et al. 2003),a number of social scientists have pursued a very diVerentcritique that uses diVerent evidence and diVering methodsof analysis. That latter critique focuses on the textual evi-dence of how scientists use racial and other social identi-ties, applying various social theories to explain why thosecategories are reproduced in genetic analyses and how thatreproduction contributes to the continued racialization ofsociety by reinforcing the historical practice of attributingbiological diVerences to racial and other social identities(Gannett 2001).

These critiques go beyond just the work of those scien-tists who explicitly endorse biological applications of racial

categories to include also the work of many more scientistswho would not endorse that view. In eVect, the social cri-tique engages in a form of literary analysis that “reads” raceeven when the use of racial categories is carefully qualiWedas a heuristic or disclaimed outright by researchers (Fujimuraet al. 2008). Thus, social critics tend to ignore or discountscientiWc issues such as population stratiWcation while treat-ing population designations such as geographic location,nationality, continent of origin, or ethnicity as surrogatesfor race (Happe 2006).

Advocates of this approach take the view that any racialimplication that can be read into a scientiWc design is evi-dence for the racialization of the project and its Wndings(Kahn 2006). They also tend to assume that racialization isa harm that outweighs any health-related beneWts that mightcome from using racial categories (or, for that matter, anypopulation-speciWc categories that might be read to implyrace) as a heuristic to discover medically applicable bene-Wts (Fullwiley 2007). Indeed, social critics are particularlyconcerned that genetic research using or implying racialcategories will contribute to the racialization of medicine(Lee 2005). At the same time, they often assert that geneticresearch detracts from, in their view, more beneWcialresearch into and prevention of environmental contributorsto disease and lack of drug response (Paradies et al. 2007).Much like the biological perspective it disparages, thesocial critique also reduces the relative complexity ofdiVerent kinds of population-speciWc identities and ofsocial distinctions within those categories to a smaller num-ber of racial categories, with a similar reduction in thepower of the social critiques to analyze the social implica-tions of genetic Wndings beyond asserting the blanket claimof racialization.

A shared basis for disagreement

It is clear that, for the most part, scientiWc proponents of thebiological meaning of racial and other social categories andsocial critics of the use of racial and other social categoriesin biological research are talking past one another. Socialcritiques tend to be published in social science rather thanlife science venues, predominantly cite other such critiquesrather than the scientiWc literature in question, and do notengage the scientiWc arguments involved in population-based genetic research. Indeed, the preferred form ofempirical evidence employed by social critics is to inter-view or ethnographically observe geneticists at work,focusing on the mode of scientiWc production rather thanthe product itself (i.e., scientiWc designs and Wndings). Incontrast, geneticists who have written about race tend topublish in scientiWc venues, rarely cite their social critics,and usually do not engage with social analyses of the racial,

123

358 Hum Genet (2009) 126:355–362

ancestral, geographic, and other identities they use to grouptheir data. The preferred form of empirical evidenceemployed by geneticists is to analyze population-speciWcsamples, comparing the distribution of genetic variantsamong those.

Not only do quantitative analyses such as frequency dis-tributions or analyses of population stratiWcation and quali-tative interpretations such as textual interpolations fail tointersect in any meaningful or productive way, the ultimatescientiWc metrics such as discoveries of disease and drugresponse variants or reductions in the social reproduction ofracial categories also are not measured or framed in a waythat allows scientiWc beneWts of population-speciWc geneticstudies to be weighed against the risks of racialization.Thus, there should be little surprise that the debate goes onwithout prospect for resolution. There is, though, one areain which the two perspectives do have common ground.That is in the way in which each elides smaller scale sam-ples of members of speciWc local populations into Wndingsabout much larger ethnic, racial, continental, and otherancestral categories.

Quantitative analyses of genetic variants necessarily arecarried out on the DNA of relatively small numbers of par-ticipants who often are recruited from the same local areaor aggregated from multiple local recruiting sites. Regard-less where those local recruitment sites are around theworld, the people who reside in each have quite complexancestral histories (and not always the same combination ofancestral complexities within each local community).Because of the way in which humans populated the Earth,some aspects of genetic variation in local populations willreXect larger ancestral categories (Tang et al. 2006), butthose larger categories (such as race or continent of origin)cannot explain all variation within and between each localpopulation. More importantly, perhaps, a number of geneticfeatures that vary between individuals cannot be explainedby race or continent of origin.

Yet the practice of scientiWc proponents of racial catego-ries is to treat samples from local populations as tokens oflarger ancestral groups. Using a similar logic, social criticsof the use of race in biological analyses also tend to treatany population-speciWc sampling of DNA as a surrogate forlarger racial categories, even when scientists speciWcallycaution against such a leap (Hunt and Megyesi 2008). Forboth sides engaged in the race and genetics debate, popula-tion-speciWc sampling almost always equates with race.This situation is further complicated by the large majorityof genetics researchers who are not engaged in that debatebut who cite self-reported participants’ social identitiesunreXexively, thereby implying that population-speciWcsampling for genetic analyses equates with social identity.

To some extent, the ongoing debate has burdenedanalytic categories such as “population,” “ancestry,” and

“geographic origin or ancestry” with implications of race.Geographical inferences of ancestry, for example, are likelymuch more useful heuristics for predicting genetic risk fac-tors for many health-related traits than racial categories(Bamshad and Guthery 2007; Handley et al. 2007), with thepotential for some degree of utility in epidemiologicresearch, public health disease prevention, and medicaldiagnosis. However, the observation that geographical evi-dence can be predictive of some patterns of genetic varia-tion has been used by scientiWc proponents to validate thecontinuing biological validity of racial categories toaccount for genetic and phenotypic patterns in contempo-rary populations (Halder et al. 2008) while social criticshave tried to argue that geographic ancestry whenever usedin a genetic context is little more than a euphemism for race(Fullwiley 2008), both tending to distract from the actualanalytic utility of geography, ancestry, and population.

That distraction is not necessarily just academic innature. Some of the social critics have advocated govern-mental regulations to determine “appropriate” deWnitionsfor and use of racial and ethnic identities in genetic researchand drug and diagnostic development (Lee and Mudaliar2009), which could eVectively institutionalize the unpro-ductive aspects of the current debate in policy restrictionson the scientiWc design of federally funded studies. Similarpolicy-driven guidelines issued by the U.S. OYce of Man-agement and Budget in the 1970s signiWcantly contributedto perpetuating and indeed increasing the unreXexive use ofracial and ethnic categories in biological research (Snipp2003).

The HapMap and race

The assumption that population-speciWc genetic Wndingsequate with racial categories is perhaps best exempliWed bythe way in which the two sides have interpreted the designand results of the International HapMap Project (HapMap).The goals of the HapMap were to discover a large numberof single nucleotide polymorphisms (SNPs) in which theminor allele is common (>5%) and to develop a map ofthese SNPs to determine areas in which SNPs are stronglyassociated in forming haplotype blocks. Because of thestrong associations within these blocks, a single SNP canrepresent the entire block and is called a tag SNP. A ran-dom selection of tag SNPs greatly reduces the amount ofgenotyping needed for genome-wide association studiesbecause the tags capture most of the variation that exists inother, nearby SNPs (International HapMap Consortium2003). Samples were collected from four populations (peo-ple of European ancestry living in Utah, people of Yorubaancestry living in Nigeria, people of Japanese ancestry liv-ing in Tokyo, and people of Han Chinese ancestry working

123

Hum Genet (2009) 126:355–362 359

at a university in Beijing) to reXect some of the diversity ofthree diVering population histories (European, African, andAsian), although none was assumed to be “typical” of anyone of those histories.

What is remarkable about the HapMap data is that thetag SNPs derived from those four populations have provento be portable to almost all other human populations testedto date (Need and Goldstein 2006), largely because morecommon variations tend to be old and so are shared acrosspopulations. In some cases, the tags that work best are thosefrom HapMap populations that are most closely related his-torically to the target population. In other cases, however,HapMap tags work quite well in more historically distantpopulations (Gonzalez-Neira et al. 2006; Service et al.2007). A global set of tags derived from all four HapMappopulations (with some supplemental African data toaccount for the greater genetic diversity in what were theoriginal human populations) may be the best possible mapfor comparative genomic analyses of disease and drugresponse (Conrad et al. 2006). While the portability of Hap-Map tags does not mean that genetic diversity is organizedidentically in populations in which the same tags are useful,it does demonstrate that the results of the HapMap projecthave been used in some ways that confound traditionalracial categories.

One would expect that the portability of HapMap dataacross human populations, the Wndings that some patternsof genetic variation either are universal or vary within tradi-tional racial categories, and the demonstration that mosthuman genetic variation is common in all human popula-tions (just at diVering frequencies) should greatly reducethe noise about race coming from both sides of the ongoingdebate. Social critics, however, have routinely referred tothe HapMap as a “race-based” project (Lee 2006; Reardon2007; Hamilton 2008), while scientiWc proponents of theutility of ancestral categories such as race have redoubledtheir arguments for the importance of identifying thosegenetic features that conform to traditional continent-of-origin categorizations and population-speciWc diVerences indiscovering disease susceptibility and drug response vari-ants (Risch 2006; Allocco et al. 2007).

The disconnect between HapMap Wndings and theresponse (or lack of response) to these Wndings by thosemost centrally engaged in the debate about race and genet-ics is a classic example of the way in which both scienceand social critiques of science can be driven by selectiveinference. Racial categories can be inferred to have biologi-cal meaning by selectively prioritizing those patterns ofpopulation genetic variation that come closer to Wttingthose categories, while population genetic designs and Wnd-ings can be inferred to be racialized by selectivelyextrapolating those population-speciWc demographic featuresthat come closer to implying racial categories. Interestingly,

both sides in the debate rely on Wgurative language or rea-soning to arrive at race, and once there treat that semanticdestination as literal. In this manner, haplotype data fromthree discontinuous geographic samples (the Japanese andChinese samples turned out to be continuous for purposesof identifying common SNPS) that were used to success-fully identify common variants and tags, which can befound across most if not all human populations were re-interpreted in the context of the race and genetics debate tobe all about race and racialization.

Can we move forward?

The scientiWc success of the initial HapMap analyses isbeing followed up by more systematic, localized popula-tion-speciWc sampling, thereby reducing artiWcial disconti-nuities in human genetic variation that have resulted fromsmall samples from a small number of populations. Withhigher-throughput genomic analyses, an increasingly largeamount of population-speciWc data will become available,including greater information on rare variants many ofwhich, because they are relatively recent, will be foundonly in single populations. These data will add to the ongo-ing scientiWc question of whether the global distribution ofhuman genetic variation is best analyzed by discontinuousclusters (Rosenberg et al. 2005), continuous clines (Serreand Paabo 2004), overlapping clusters (Witherspoon et al.2007), or some more complex combinations of all of theabove (Novembre and Stephens 2008).

It is likely, though, that the debate about race and geneticswill simplify the sophisticated detail and complexity ofthese new data and analyses by using the same reductionistracialized logic. After all, that is how the debate has man-aged to continue to reproduce itself over all the precedingyears. If, however, we step away from those closed systemsof argumentation, we can gain a diVerent perspective on theso far unproductive debate.

One path forward is to treat the general question of therelationship between social identities and genetic Wndingsnot as a debate but, rather, as a research question thatrequires both scientiWc and social scientiWc investigationand then the integration of those scientiWc and social scien-tiWc Wndings. Thus, the challenge is to generate testablehypotheses, empirical data relevant to those hypotheses,and more precise analyses of those data. This is a challengefor both sides of the current debate as well as for geneticsresearchers who have not engaged in the race and geneticsdialog but who have used social identities in an unreXectivemanner.

The challenge for scientiWc proponents of the geneticsigniWcance of racial and other social categories is toanalyze “genetic ancestry” in terms of the complex

123

360 Hum Genet (2009) 126:355–362

associations among multiple markers used for populationstratiWcation. Analyses of those marker associations willproduce multiple, alternative genetically based deWnitionsof populations (depending on where thresholds for homo-geneity are set) that do not depend on racial and other sociallabels and that are heuristic constructions for the explicitpurpose of facilitating association and other studies. Whileone could then compare various heuristic deWnitions of“genetic ancestry” with participant self-reports of socialidentity, doing so would be extrinsic to a study’s samplingand analysis designs. Comparisons of genetic constructionsof ancestry and participant self-reports of social identitywould have little scientiWc value as doing so greatlyreduces the explanatory as well as translational power ofthe more detailed genetic constructs. It would be importantto avoid using existing social labels for the analytic con-structions of genetic ancestry so as to avoid reifying themas biological deWnitions of social identity.

The challenge for social critics is to specify the prospec-tive risk of racialization in speciWc examples in ways thatcan be investigated and either validated or falsiWed. Currentclaims of racialization cannot be tested, one way or theother. Testable population-speciWc risks of racializationthen can be weighed against potential health and other sci-entiWc beneWts from using those social identities to recruitand/or identify groups of participants in particular geneticstudies. The plausibility of “reading race” from scientiWcdesigns and publications is very diVerent from demonstrat-ing the actual risk or eVects of racialization. While the latterwould require some development of instrumentation andthen empirical investigation, social scientists already havedeveloped validated measures of the eVects of racial dis-crimination that could be adapted to the purpose (Krieger2000; Blanke et al. 2004).

The challenge for any genetics researchers who usesocial identities to recruit study participants and/or to reportstudy Wndings is to treat those identities as analytic catego-ries the criteria for which must be explicitly speciWed andrigorously applied. Thus, social identities used in geneticstudies should be explicitly qualiWed with respect to (1) thecriteria used to deWne those identities by the researcher forpurposes of the study (although those same criteria may ormay not be the ones used by participants to identify them-selves), (2) the extent to which local samples are represen-tative of much larger populations, (3) the social processesby which non-biological events and forces have shaped his-torical and contemporary claims to those identities, and (4)the underlying scientiWc reasons for using those identitiesin a biological context. Treating social identities used ingenetic research as analytic categories will both reduce theextent to which they are used unreXexively in such studiesas well as explicitly qualify the extent to which particular

genetic Wndings are associated with speciWc social identi-ties.

Being able to specify and weigh analyses of both the sci-entiWc and social evidence about the relationship betweengenetics and social identities—rather than reducing both toclaims about race—gives us a good chance of actuallyresolving many of the outstanding issues in the longstand-ing debate, as doing so requires us to have a practical,empirically based discussion about the trade-oVs in howrace and other social identities are used in biological con-texts. We can weigh the heuristic value that race or othersocial categories might have in answering a speciWcresearch question with the implications that biological usesof those racial and other social identities might have inlarger social and political contexts. In some cases, forexample, a racial category may be unavoidable because ofthe ways in which historical demographic, epidemiologic,and other data necessary for the analysis of linked pheno-typic outcomes have been collected over the years. In othercases, though, racial and other social identities may be usedsimply because they are available as data points, withoutmore considered scientiWc consideration of their accuracyor utility, and so should be discarded.

Certainly, diVerent people can and will argue for diVer-ent ways of analytically specifying, measuring, and weigh-ing potential scientiWc beneWts and potential social harms inusing race and other social identities in biological contexts.The important thing, though, is to focus on empirical scien-tiWc and social scientiWc analyses of those potential beneWtsand risks in which the two can be weighed together,because then there is some chance of a resolution (at leaston a case by case basis). In some instances, that resolutionmay consist of using social identities as convenient heuris-tics to get at patterns of genetic variation, but then alsousing social analyses to make clear why racial and othersocial categories are being used and what should not beinferred from that use.

To a large extent, the current version of the race andgenetics debate has been powered by the unreXexive use ofsocial identities by scientists and the unreXexive use ofracialization by social critics. By rendering those twoimpulses reXexive, most of what has been argued in therace and genetics debate to date recedes in the face ofempirically grounded analyses. What remains is the muchmore complex proposition of relating biological processeswith social processes in a sophisticated manner, which isboth the promise and the problem of gene–environmentinteraction (including epigenetics). Arguably, the unpro-ductive focus on debates about race and genetics, and itsspillover onto population-speciWc studies, has distracted usfrom more detailed measurements and analyses of social–biological interactions.

123

Hum Genet (2009) 126:355–362 361

The absolutist proponents and opponents of using racialcategories in genetic research are both stuck in a racializedway of thinking. They are looking for race in all the wrongplaces. By privileging race (albeit for diVering purposes),they each have lost touch with goals that should not bemutually exclusive, but that they have collaborated in mak-ing seem to be so: using population genetics to reduce theadverse eVects of disease susceptibility and drug responsevariants and using social critiques to reduce the adverseaVects of racialization. We should not assume, as the abso-lutists on each side have done, that it is impossible toaccomplish both these goals at the same time.

Acknowledgments This paper beneWted from discussions at theBanbury Center conference, “Who are we? Kinship, ancestry, andsocial identity,” which was funded in part by the Richard LounsberyFoundation.

References

Allocco DJ, Song Q, Gibbons GH, Ramoni MF, Kohane I (2007)Geography and genography: prediction of continental origin us-ing randomly selected single nucleotide polymorphisms. BMCGenomics 8:68. doi:10.1186/1471-2164-8-68

Bamshad M, Guthery SL (2007) Race, genetics and medicine: does thecolor of a leopard’s spots matter? Curr Opin Pediatr 19:613–618

Blank RM, Dabady M, Citro CF (eds) (2004) Measuring racial dis-crimination: panel on methods for assessing discrimination. Na-tional Academies Press, Washington, DC

Boas F (1912) Changes in bodily form of descendants of immigrants.Am Anthropol 14:53–562

Burchard EG, Ziv E, Coyle N, Gomez SL, Tang H, Karter AJ, Moun-tain JL, Perez-Stable EJ, Sheppard D, Risch N (2003) The impor-tance of race and ethnic background in biomedical research andclinical practice. N Engl J Med 348(12):1170–1175

Burchard EG, Borrell LN, Choudhry S, Naqvi M, Tsai HJ, Rodriguez-Santana JR, Chapela R, Rogers SD, Mei R, Rodriguez-Cintron W,Arena JF, Kittles R, Perez-Stable EJ, Ziv E, Risch N (2005)Latino populations: a unique opportunity for the study of race,genetics, and social environment in epidemiological research. AmJ Public Health 95(12):2161–2168

Campbell CD, Ogburn EL, Lunetta KL et al (2005) DemonstratingstratiWcation in a European American population. Nat Genet37(8):868–872

Chakravarti A (1999) Population genetics—making sense out ofsequence. Nat Genet 21(1s):56–60

Choudhry S, Burchard EG, Borrell LN, Tang H, Gomez I, Naqvi M,Nazario S, Torres A, Casal J, Martinez-Cruzado JC, Ziv E, AvilaPC, Rogriquez-Cintron W, Risch NJ (2006) Ancestry-environ-ment interactions and asthma risk among Puerto Ricans. Am JRespir Crit Care Med 174(10):1086–1091

Conrad DF, Jakobsson M, Coop G, Wen XQ, Wall JD, Rosenberg NA,Pritchard JK (2006) A worldwide survey of haplotype variationand linkage disequilibrium in the human genome. Nat Genet38(11):1251–1260

Cooper RS, Kaufman JS, Ward R (2003) Race and genomics. N EnglJ Med 348(12):1166–1170

Fujimura JH, Duster T, Rajagopalan R (2008) Introduction: race,genetics, and disease: questions of evidence, matters of conse-quence. Soc Stud Sci 38(5):643–656

Fullwiley D (2007) Race and genetics: attempts to deWne the relation-ship. Biosocieties 2:221–237

Fullwiley D (2008) The biologistical construction of race: “admixture”technology and the new genetic medicine. Soc Stud Sci 38:695–735

Gannett L (2001) Racism and human genome diversity research: theethical limits of “population thinking”. Philos Sci 68(3):S479–S492

Gonzalez-Neira A, Ke XY, Lao O, Calafell F, Navarro A, Comas D,Cann H, Bumpstead S, Ghori J, Hunt S, Deloukas P, Dunham I,Cardon LR, Bertranpetit J (2006) The portability of tagSNPs acrosspopulations: a worldwide survey. Genome Res 16(3):323–330

Goodman AH (2000) Why genes don’t count (for racial diVerences inhealth). Am J Public Health 90(11):1699–1702

Halder I, Shirver M, Thomas M, Fernandez JR, Frudakis T (2008) Apanel of ancestry informative markers for estimating individualbiogeographical ancestry and admixture from four continents:utility and applications. Hum Mutat 29(5):648–658

Hamilton JA (2008) Revitalizing diVerence in the HapMap: race andcontemporary human genetic variation research. J Law Med Eth-ics 36(3):471–477

Handley LJ, Manica A, Goudet J, Balloux F (2007) Going the distance:human population genetics in a clinal world. Trends Genet23(9):432–439

Happe KE (2006) The rhetoric of race in breast cancer research. Pat-terns Prejudice 40(4–5):461–480

Hunt LM, Megyesi MS (2008) Genes, race and research ethics: who’sminding the store? J Med Ethics 34(6):495–500

International Congress of Eugenics (1923) ScientiWc papers of the sec-ond International Congress of Eugenics: held at American Muse-um of Natural History, New York, 22–28 September 1921,Committee on Publication, Charles B. Davenport, Chairman.Williams & Williams, Baltimore

Kahn J (2006) Genes, race, and population: avoiding a collision of cat-egories. Am J Public Health 96(11):1965–1970

Klimentidis YC, Miller GF, Shriver MD (2009) Genetic admixture,self-reported ethnicity, self-estimated admixture, and skin pig-mentation among hispanics and native Americans. Am J PhysAnthropol 138(4):375–383

Krieger N (2000) ReWguring “race”: epidemiology, racialized biology,and biological expressions of race relations. Int J Health Serv30(1):211–216

Lee SSJ (2005) Ethical implications of race and genomics—racializingdrug design: implications of pharmacogenomics. Am J PublicHealth 95(12):2133–2138

Lee SSJ (2006) Biobanks of a ‘racial kind’: mining for diVerence in thenew genetics. Patterns Prejudice 40(4–5):443–460

Lee SSJ, Mudaliar A (2009) Racing forward: the Genomics and Per-sonalized Medicine Act. Science 323(5912):342

Lewontin RC (1974) The genetic basis of evolutionary change. Colum-bia University Press, New York

Marchini J, Cardon LR, Phillips MS, Donnelly P (2004) The eVects ofhuman population structure on large genetic association studies.Nat Genet 36(5):512–517

Need AC, Goldstein DB (2006) Genome-wide tagging for everyone.Nat Genet 38(11):1227–1228

Novembre J, Stephens M (2008) Interpreting principal componentanalyses of spatial population genetic variation. Nat Genet60:646–649

Paradies YC, Montoya MJ, Fullerton SM (2007) Racialized geneticsand the study of complex diseases—the thrifty genotype revisited.Perspect Biol Med 50(2):203–227

Reardon J (2007) Democratic Mis-haps: the problem of democratiza-tion in a time of biopolitics. Biosocieties 2:239–256

Risch N (2006) Dissecting racial and ethnic diVerences. N Engl J Med354(4):408–411

123

362 Hum Genet (2009) 126:355–362

Risch N, Burchard E, Ziv E, Tang H (2002) Categorization of humansin biomedical research: genes, race and disease. Genome Biol3:comment2007.1-2007.12

Rosenberg NA, Mahajan S, Ramachandran S, Zhao C, Pritchard JK,Feldman MW (2005) Clines, clusters, and the eVect of study de-sign on the inference of human population structure. PLoS Genet1:e70

Sankar P, Cho MK, Mountain J (2007) Race and ethnicity in geneticresearch. Am J Med Genet A 143A(9):961–970

Serre D, Paabo SP (2004) Evidence for gradients of human geneticdiversity within and among continents. Genome Res 14(9):1679–1685

Service S, Sabatti C, Freimer N (2007) Tag SNPs chosen from Hap-Map perform well in several population isolates. Genet Epidemiol31(3):189–194

Snipp CM (2003) Racial measurement in the American census: past prac-tices and implications for the future. Ann Rev Sociol 29:563–588

Tang H, Quertermous T, Rodriguez B, Kardia SLR, Zhu XF, Brown A,Pankow JS, Province MA, Hunt SC, Boerwinkle E, Schork NJ,Risch NJ (2005) Genetic structure, self-identiWed race/ethnicity,and confounding in case-control association studies. Am J HumGenet 76(2):268–275

Tang H, Coram M, Wang P et al (2006) Reconstructing genetic ances-try blocks in admixed individuals. Am J Hum Genet 79(1):1–12

Tian C, Gregersen PK, Seldin MF (2008) Accounting for ancestry:population substructure and genome-wide association studies.Hum Mol Genet 17(2):R143–R150

Witherspoon DJ, Wooding S, Rogers AR et al (2007) Genetic similar-ities within and between human populations. Genetics176(1):351–359

Yang JJ, Burchard EG, Choudhry S et al (2008) DiVerences in allergicsensitization by self-reported race and genetic ancestry. J AllergyClin Immunol 122(4):820–827

123