Lumpers and Splitters, Individuals and Structures


Hochschild JL. Lumpers and Splitters, Individuals and Structures. In: Racialized Politics: The Debate about Racism in America. edited by David Sears, Jim Sidanius, and Lawrence Bobo. Chicago IL: University of Chicago Press ; 2000. pp. 324-343.

Full Text


Lumpers and Splitters, Individuals and Structures


Jennifer L. Hochschild

Department of Politics

Princeton University 

December 29, 1998 



In Racialized Politics: The Debate about Racism in America, ed. David Sears, James Sidanius, and Lawrence Bobo (University of Chicago Press, forthcoming 2000)


NOTE: not quite final version.





            My initial reaction upon reading these papers, in conjunction with other recent writings of these scholars, was the same as it has been for a decade or more: I was convinced by each in turn that proponents of principled conservatism, symbolic racism, and social structural approaches were correctly analyzing white Americans’ racial views.  But according to all three sets of authors, the arguments of the other two sets were less correct than their own, if not simply wrong.  It is not possible, of course, for all three to be correct if they set their arguments up in opposition to each others’.  Yet each set of proponents presents elegant theoretical arguments, strong evidence, sophisticated methodological dissections of that evidence, and moral commitments to racial justice and social scientific truth. Furthermore, in many cases they are analyzing precisely the same data sets. What is going on?

            I see two kinds of explanations for my (and others’?) inability or unwillingness to accept one of these approaches at the expense of the other two.  The first cluster of explanations lies within the paradigm of survey research in which these authors work; the second explanation puts that paradigm into a broader structural context.  My goal in these comments is not to come down in favor of one or another approach but rather to explore the commitments one makes by accepting one or another of them.  I will also argue for approaches that move outside all three of the positions featured in Racialized Politics, and conclude with a sketch of my own view of the puzzles presented by racialized politics in the United States.


Explanations of Racial Politics within the Survey Research Paradigm

Lumpers and Splitters: For several centuries taxonomists of the natural world have been divided into lumpers, who seek to merge a larger number of proposed species or genera into a smaller number, and splitters, who seek to move in the opposite direction.  Some of the debate among the three basic approaches on display in Racialized Politics has the flavor of this difference in taste (or in epistemological starting point, to be more formal). 

Sniderman and his colleagues believe that one must separate out characteristics of individuals that are often associated with racial animus from that racial animus itself in order to determine how much effect prejudice has on political views.  Thus they carefully distinguish values such as individualism or egalitarianism, ideologies such as political conservatism or liberalism, levels of education, and so on from negative racial affect. Once the affect is isolated, they argue, one can examine its influence on policy views and other dependent variables, either alone or in conjunction with some of the characteristics from which it was first isolated.  But only when the concept of racial prejudice is purified from other concepts can we see what real work it does in individuals’ psyches and political stances.  As they put it, “levels of support for government assurance of equal opportunity to succeed for a group…[are] a function of three factors – the group to benefit, blacks or women [that is, prejudice]; respondents’ political point of view, liberal or conservative [that is, political ideology or principles]; and … formal education” (draft, p. 25).[1]

Kinder and his colleagues, in contrast, believe that splitting more and more characteristics away from racial prejudice artificially shrivels it into irrelevance-by-definition; racial affect must be lumped together with certain other values and emotions with which it is closely associated in order to see its real force. Those values and emotions may well change over time, so that racism in the 1990s involves a slightly different set of elements than racism did in the 1950s.  Thus “for reasons of history, individualism and prejudice have become entangled: properly conceived, individualism is in prejudice” (Kinder and Mendelberg, draft p. 1). Sears and Henry (this volume, p. 37 draft) concur, concluding after examination that “symbolic racism originat[ed]… in a racialized mixture of individualistic and inegalitarian values.”[2]

            As with a taxonomy of species and genera, this debate seems to me more a matter of preference, or of usefulness for a particular purpose, than of correct or incorrect theorizing.  If one narrows the definition of racial prejudice, one is likely to find less of it or to find that it is less influential in comparison with other motivators than if one includes those motivators within one’s definition of racial prejudice.  That is a matter of simple arithmetic logic.  Whether one should narrow the definition of racial prejudice in this way, or conversely broaden it to include various other values and beliefs, is a matter of theoretical purpose.  One should narrow it if the question is how and how much whites’ feelings about African Americans, as distinguished from other feelings and values, affect what they think should be done by or for African Americans.  One should broaden it if the question is whether and why whites react differently to African Americans than to other groups, or differently to policy issues that are racially-inflected than to policy issues that are not.  They are different questions; they are both legitimate.

            Choosing to be a lumper or a splitter also has political implications, if not political motivations.  Splitters are likely to give comfort, whether intentionally or not, to those who see racial divisions as simply one among many equally important social differences in a complex society.  Lumpers are likely to give comfort, whether intentionally or not, to those who see racial divisions as more important than, or even causing, most other social differences.  Much of the animus in this research arena derives, so far as I can tell, from normative disputes between those who see racial inequality as the central issue and study public opinion in order to alleviate it, and those who see the role of public opinion in governance as the central issue and study racial attitudes in order to understand that link. 

            Of course, one can go too far in lumping or splitting, as none of the authors in this volume do, to a point where the enterprise loses intellectual legitimacy.  Dinesh d’Souza defines racism very narrowly – as a belief in the biological inferiority of the subordinated race[3] – and thereby can claim that illegitimate racism has practically disappeared in the United States.  Conversely, some proponents of affirmative action claim that any opposition to even its strongest forms is evidence of racism.[4]   Both arguments are wrong, largely because most of the work is done by choosing a particular definition of “racism.”  Thus it is not silly to worry about whether an author is taking everything meaningful out of a concept or is stuffing too much into it.  But within the range of arguments represented by the authors in Racialized Politics and the writers to which they are responding, different operationalizations of key terms are appropriate to different research questions and different normative starting points.


Figure and Ground:  A different metaphor from a different craft also helps to explain why each approach seems convincing on its own terms, yet all insist that they cannot all be correct.  What is a central theoretical or empirical concern to one set of authors often is a relatively trivial or unproblematic fragment of the analysis to another.  One must, of course, simplify some questions in order to focus detailed attention on others, but what one chooses to set in the center of the picture or relegate to the background can be (and here is) subject to dispute.

            Consider the concept of affirmative action, for example.  Sidanius and his colleagues measure “respondent reactions to a single item: ‘affirmative action.’” (this volume, draft p. 19 [they do not provide question wording]).  Sniderman and his colleagues have often relied heavily on the series of National Election Studies (NES) questions, which include the word “preference” for the query about jobs and (in some years) the word “quota” for the query about college admissions.[5]  Their 1986 Race and Politics Survey (Sniderman and Piazza 1993) also uses the word “quota” in the question about admissions, and the phrase “law to ensure that a certain number of federal contracts go to minority contractors” in the question about set-asides.  Their 1991 Race and Politics Study (Sniderman and Carmines 1997) asks if “companies should be required to give a certain number of jobs to blacks.”[6] 

And yet Larry Bobo (this volume, draft p. 2), building on the work of Steeh and Krysan (1996) and his own previous work, correctly castigates “the distorted view that opposition to affirmative action among whites is monolithic.”  It is not just that whites oppose some forms of affirmative action less than they oppose other forms.  In fact, they consistently support some forms, such as educational outreach programs or job training and outreach programs.  Furthermore, although most whites vehemently oppose “quotas” and “preferences,” so do many (sometimes most) blacks, Latinos, and Asians. So do the courts; except in very specific circumstances, quotas have been held to be illegal. In fact,  “quotas are a rarity in all affirmative action programs focused on employers” (Reskin 1998: 5).  Thus to ask survey respondents about “support for affirmative action” (as in Sidanius et al. and in Sniderman et al., both in this volume) is to leave indeterminate what the respondent is in fact responding to.  Similarly, to assume (as in Sniderman’s, Sears’, and Kinder’s work) that one has captured views on affirmative action through responses to questions referring to  “preferences” or “quotas” is probably to exaggerate the amount of opposition and possibly to distort the relationship between political ideology and views of affirmative action.[7]

            Whether a more careful treatment of the concept of affirmative action would change Sniderman and his colleagues’ – or Kinder and his colleagues’ – results, or lead either group to reformulate their theoretical approach, remains unknown.  It might not, but I can imagine how it could.  In the list experiment, for example, Sniderman et al. might find less covert anger at “black leaders asking for educational outreach and job training for blacks” than at “black leaders asking for affirmative action.” Would those people who are less angry as a consequence of the new wording be more liberal? More conservative? More individualistic?… We do not know, but several hypotheses are plausible and it would be worthwhile to find out which, if any, obtains.  Similarly, in the section of the chapter which analyzes liberals’ and conservatives’ views of governmental intervention, the relationship between political ideology and racial prejudice might be different if affirmative action were defined as “outreach and training programs” rather than as preferences and quotas.  That new result could, in turn, significantly affect the conclusions in their core argument about the importance of the specific political choices that citizens face.

Kinder and his colleagues have a slightly different problem of figure and ground in their treatment of individualism.  As Sniderman and others have pointed out, most of the canonical measures of individualism (in the NES) invoke the idea of hard work.  That set of wording choices probably says a lot about academics who write survey questions.  But it does not necessarily say a lot about individualism – which could encompass a sense of individual autonomy, pride in one’s uniqueness, fear of entangling alliances, political independence, rejection of ascriptive labeling, or other conceptions that have little to do with hard work.   Might measures of individualism have more effect on racial politics if they were operationalized more broadly or differently?  And if so, would that change Kinder’s claim that individualism has no effect on racial policy views because “individualism is in prejudice”?  We do not know, because Kinder and other authors of the NES have seen the individual–ness of the NES questions as the figure, and the hard work connotations as the background.[8]

            For a third example, consider the value of equality.  Many writers within the tradition of survey research have treated the concept of egalitarianism as relatively simple and single-dimensional.  But, as Sears and his colleagues show in this volume, it is not.  They report on one distinction within the global concept of egalitarianism – “consensual support” for “the general principle of equality” for both individuals and groups versus the “highly contested…assertion that our society continues to have insufficient equality and repairing that flaw would improve the society” (draft, p. 24).  That distinction helps to bring the vague background more sharply into focus as the central figure with texture and shadings.  But in making that distinction clearer, they allow other meanings of equality, at least as important, to remain muddled. 

Equality means many different things, some of which appear distinctly inegalitarian to proponents of some other definition of equality.[9]  The simplest division is between equal opportunity and equal results -- most people prefer one to the other, and it seems pointless to assert that one or the other is the “real” egalitarian preference.  Furthermore, the NES/LACSS egalitarianism items are ambiguous even if one is prepared to assert that a particular understanding of equality is the “real” one.  Consider “If people [or “blacks and whites” or “men and women”] were treated more equally in this country we would have many fewer problems.”   I could agree with that as a strong proponent of equal opportunity, or as a strong proponent of equal results. (After all, the term “treated equally” can be understood to mean “giving the same treatment” as in the same medicine or the same amount of schooling, or as “treated with the same amount of respect,” or perhaps as “given the same outcome.”) It is impossible, therefore, to know if the contestation that Sears et al. report over this question results from citizens’ real disagreement about the value of equal results, or from their real disagreement about the value of strong forms of equal opportunity, or from varying interpretations among them about what the question is asking.  And yet Sears and his colleagues interpret the contestation as evidence that the questions “force tradeoffs between egalitarian and individualistic values;” they interpret only the items on which most Americans agree to be “egalitarian” (draft, p. 26).  Neither philosophically nor empirically does that conclusion seem warranted at this point.

            One could continue this discussion for a long list of concepts used in these studies.  For example, how settled are social groups in Sidanius et al.’s study, given that 11% of their student respondents identified themselves as “other” rather than as white, black, Hispanic, or Asian?[10]  If a substantial fraction of respondents are unwilling to classify themselves racially or ethnically, might they also envision very different entities, either within a single person’s set of survey responses or across respondents, when they consider undefined “groups” in the 16-item SDO index?  And might these fluid and changeable understandings of “groups” affect the robustness of the underlying theory about social dominance of one group over others?  We do not know.

            As before, my point is not to specify which concepts should be the figure and which should be relegated to the background.  Also as before, the “right” choice depends on what one most urgently needs to know to answer a particular question.  Sniderman does not need to worry as much as Sidanius et al. about what respondents mean by a social group, whether in their self-identification or in their answers to survey questions.  But Sniderman and his colleagues (as well as Sears et al., and Kinder et al.) should worry much more than they have up to this point, and probably more than Sidanius and his colleagues need to worry, about the various meanings of affirmative action.  That issue, after all, is crucial to understanding how liberal and conservative ideologies function in racial policy preferences.

There is a broad silver lining to this cloud.  My claim about the need to bring vague background concepts into the sharply etched foreground gains credence from the results of what these scholars have discovered by beginning to do precisely that.  Researchers realized only recently that egalitarianism has much more impact on racial policy views than does individualism, at least as both values are commonly measured.  Given that newly-clear pattern, it is now more urgent both to deconstruct the various meanings of equality and to broaden the operational definitions of individualism.  Similarly, only within the past few years have enough surveys cumulated on evaluations of different affirmative action practices for us to realize just how systematic and meaningful are the public’s varied responses to it.   New research on the social construction of group identity (as well as the social fact of an increasing number of mixed-race children) is sensitizing us to the permeability of group boundaries and the fluidity of individual self-identification.

This discussion could be formulated as a variation on the theme of lumping and splitting. Disparate meanings of concepts such as affirmative action or equality are lumped together when the concept remains in the background, and are split apart when the concept is brought into the glaring spotlight of the foreground.  Unlike with the first form of lumping and splitting, in this case I have a starting premise: the concepts discussed here, among others, need to be treated to a more subtle array of measures and more sophisticated philosophical discussion before one can make broad claims about their causes and effects.

Many analysts agree; that proposition is insufficiently implemented more often for practical than for substantive reasons.  Whether one can be sufficiently subtle in one’s treatment of egalitarianism, individualism, affirmative action, group identity, and other concepts in the same piece of work that is already seeking to parse meanings of racism and race-related policy preferences, without drowning in complexity, remains to be seen. Probably one cannot, given the evidence of the analyses already before us.  That suggests the need for intellectual approaches beyond survey research to parse what people mean when they identify themselves as individualistic, supportive of equality, hesitant about affirmative action, or as a member of an “other” racial or ethnic group. 

One candidate is focus groups, which could ask people to discuss what they mean by these and other concepts and how they use them to make policy choices.  Some research is beginning to integrate results of focus groups with results of survey research, or to blend the two techniques in order to attain the richness of the former along with the generalizability of the latter (for example, Sigel 1996; Lamont 1992; Conover and Searing 1998; Conover, Searing, and Crewe 1998; Fried 1997; Hibbing and Theiss-Morse 1995).  Another strategy is intensive interviews, either with randomly selected individuals across an array of social locations or with people specifically selected because of their particular social location.  Here too, interviews could be combined with surveys to get the best of both worlds (e.g. Krysan 1995; Stenner 199X; Hochschild 1981).  Yet another strategy is to follow the path blazed by Sniderman and his colleagues and make surveys themselves much more interactive and attentive to individual nuances (see also Keyes 19XXX; Kinder and Sanders 1996).  Or one could do more imaginative work with open-ended responses embedded in conventional surveys (Knight 1998).

            In short, these authors appropriately criticize each other for not paying sufficient attention to the many meanings of concepts that they have found to be crucial to their own research.  They might instead, or in addition, devote their energy to developing new ways to bring more of these concepts into the foreground where they can be carefully analyzed without drowning themselves and their readers in endless detail.


Either/Or or Both/And:  We see another epistemological choice in the way that authors in Racialized Politics handle alternative hypotheses.  None reject out of hand the legitimacy of competitors to their favorite argument.  How could they? -- all of the research traditions are too robust to be completely dismissed by a responsible analyst.  But some of the authors in Racialized Politics admit the partial correctness of competing views mainly in order to show thereafter the superiority of their preferred view, whereas others set out to see how many views can plausibly be supported under specified conditions.

            Compare, for example, the strategies of Sidanius et al. and Pettigrew.  The former seek to “compare the relative adequacy of the principled-politics and social dominance models of public opposition to race-targeted policies.”  They find “two major points on which both principled politics and social dominance theorists have always agreed,” but conclude that “our results are still fundamentally at odds with essentially all of the other core predictions of the principled-politics perspective” (this volume, draft pp. 40-41, emphasis in original). Similarly, despite “a few places in which social dominance theory and symbolic racism theory take complementary positions…, the two models are still fundamentally different” (draft, pp. 47-48).

Pettigrew, in contrast, seeks “to systematize the array of correlates that persistently arise in the huge prejudice literature” in order to move “toward broader theories of individual differences in prejudice” (this volume, draft p. 3).[11]   Thus he spends most of his chapter examining as many possible explanations for prejudice as he can find, across as many nations and objects of prejudice as are available.  He does not seek to show that one explanation (or a small set) works better than others, but rather to winnow all the plausible explanations down to a manageable set that covers as much conceptual territory as possible. 

            Most contributors to Racialized Politics organize their work in terms of  “either/or” rather than “both/and” because that is how the debate among symbolic racism, principled conservatism, and social structural views has been constructed.  In my view, that logical structure is unfortunate.[12]  The arena of racial and ethnic politics is so complex, and survey research (like other methods) is so far from being able to capture all racial attitudes, that the effort to find a single best explanation for racial views seems not only impossible but misguided.[13]  As I read the evidence on the recent history of Americans’ racial attitudes, most citizens do not have simple or settled affects, perceptions, or policy preferences with regard to race that lead them actually to exclude other logically antithetical affects, perceptions, or preferences. John Zaller’s (1992) portrayal of people who carry around a basket of views and feelings which are sampled (randomly or systematically) to produce a given survey response seems more appropriate in the intricate world of racial politics than any of the fixed theoretical constructs we are offered in Racialized Politics.  Alternatively, Robert Lane’s (1962) concept of morselization – in which people hold distinct and even contradictory views in separate “parts” of their mind – suggests that respondents could be simultaneously principled conservatives (or liberals) and symbolic racists.  Alternatively, people are ambivalent about racial matters (Hochschild 1993; Smelser 1998), so they knowingly hold contradictory views.  At a minimum, the both/and construction provides some sort of resolution to the dilemma with which I began these reflections – the fact that I find each of these theories about equally persuasive.

            I am not claiming that there is no structure to American racial politics – far from it, as I shall argue in the next section.  I am claiming that white Americans’ racial views are so multifaceted or indeterminate that a “both/and” strategy for making sense of them is a more promising premise on which to build a research program than is an “either/or” strategy.


A Bifocal or Multifocal Paradigm:  Some contributors to Racialized Politics focus exclusively on the relationship between European Americans and African Americans (or between whites and minorities).[14]  That is the case for Sears et al., Kinder and Mendelberg, Sidanius et al., and Taylor. (It is also the case for Sniderman and his various colleagues; in this, at least, most contenders among the three basic approaches concur.)  For some such as Sniderman it is an unspoken starting premise of their research.  For others, it is both a starting premise and a research result.  Sidanius and his colleagues, for example, find that “SDO … made a relatively strong contribution to affirmative action opposition with respect to Blacks…. However, SDO was not significantly related to affirmative action opposition for women or the poor” (this volume, draft p 36).  This finding accords with their theoretical expectations since, “in America, Blacks will face higher levels of discrimination than other ethnic groups and will inspire the greatest degree of White opposition to policies designed to help them” (draft p. 47).

             Kinder and Mendelberg also focus exclusively on the binary relationship between blacks and whites because of both their reading of American history and their empirical findings.  They find that “the impact of prejudice is most pronounced on policies that deal explicitly and unambiguously with race…; modest though still sizeable on what might be called ‘covert’ racial issues…; and vanishes altogether on broad social programs” (this volume, draft pp. 30-31).[15]  Kinder furthermore agrees with Sidanius et al. that “divisions by race are nothing new to American politics, but if anything, they are more prominent now than they were a generation ago” (Kinder and Sanders 1996: 288).  Sears and his co-authors are perhaps a touch more optimistic (they do not argue that racism is worse than it was a generation ago), but they too keep their focus on the fact that “blacks and whites remain severely polarized over racial issues” (this volume, draft p. 3).

 Sniderman and his colleagues see greater declines in racial prejudice and its effects than do the other scholars discussed here. As they put it, “a quarter century ago, what counted was who a policy would benefit, blacks or whites; now, what counts as much, or more, is what the policy aims to accomplish and how it proposes to go about accomplishing it”(Sniderman and Piazza 1993: 4-5).[16]  But they agree with the others that “race has not receded into the background of American life. On the contrary,…race remains as divisive as ever” (Sniderman and Carmines 1997: 1)  And more importantly at the moment, by “race” they share the bifocal attention to the relationship between white and black Americans.[17]  Finally, Marylee Taylor also focuses on the relationship between whites and blacks in her analysis for this volume.[18]  She seeks to “assess the impact of black population share on racial policy opinion, racial policy-related beliefs, [and] ostensibly non-racial views often taken as alternative explanations of racial policy opinion”– by which she means whites’ views about blacks and about policies that directly or disproportionately affect blacks (this volume, draft p. 4).

In contrast with these scholars, some contributors to Racialized Politics have a multifocal framework.  Pettigrew eschews any discussion at all of white and black Americans – in fact, of Americans of any race or ethnicity.  His purpose is to understand the causes of prejudice among people of many nationalities, aimed at people of many other nationalities.  Explanations that are specific to one nation’s political ideology (e.g. the American Creed) or one nation’s distinctive outgroup (black slaves rather than Native Americans or non-Anglo immigrants) do not fit into his search for a systematic or broad theory of prejudice.  Thus he focuses on the Eurobarometer, which surveyed people of seven nationalities in four western European nations and asked them about six “target groups.”  He devotes most of his attention to analyses that hold across all of the nations, all of the nationalities, and all of the disparaged groups.  He concludes with a note of encouragement to scholars (if not to activists fighting racist practices): “mini-theories of prejudice… can be parsimoniously classified under a few basic families of explanation…. Many of the basic phenomena – political, psychological, and sociological – are shared across nations and groups” (draft, pp. 21-22).

Unfortunately, the Eurobarometer does not include a survey of Americans (who are not, after all, Europeans). An obvious next step in Pettigrew’s research program would be to see if white American prejudice against blacks or immigrants analytically resembles French, British, or German prejudice against Turks or the Roma.  If not, that leaves space for arguments about American exceptionalism, the unique effects of slavery, and so on.  If so, studies (including my own [Hochschild 1995]) that are entirely focused on the United States could usefully be reformulated in a more broadly comparative framework.  In short, we have here another case of lumping or splitting – in this case of nations or nationalities rather than of concepts or theoretical frameworks. Most American scholars of race split the American case away from other possibly analogous cases; it would be worthwhile to devote more attention to the implications of lumping the American case into a comparative framework in order to see what is in fact distinctive or shared across nations (as in Gilroy 1993; Fredrickson 1995; Marx 1998; Parikh 1997).

Bobo, and Hughes and Tuch, are multifocal in a different way.  Like most of the other authors in Racialized Politics, they focus on the United States and they include within their analyses factors that are specific to American history and practice.  But unlike most of the others they start from the premise that it is “unfortunate” that “nearly all past research on racial and ethnic differences… has focused exclusively on whites and blacks” (Hughes and Tuch, this volume, draft p. 1).  Bobo’s goals include a systematic (and rare) comparison of attitudes among blacks, whites, Latinos, and Asian Americans on affirmative action for blacks.  Hughes and Tuch carry the multifocal framework a step farther by looking at variations in all four groups’ explanations for poverty among the three minority groups. 

At this point, the story gets very complicated, in a way that should yield research agendas for generations of graduate students to come.  Bobo shows that racial and ethnic groups differ in their support for affirmative action for blacks; that result opens up the question of why Latinos (who are likely to be “white”) resemble blacks in their views on this issue, whereas Asians (who are racially not “white”) resemble whites in their views.  If Bobo’s pattern holds across other policy domains, (as it sometimes does [Hochschild and Rogers 1998]), then we can no longer talk about racial division in the United States as we have been wont to do.  It appears that two distinct races, one of which is disproportionately comprised of recent immigrants (i.e. whites and Asians), resemble each other; a third race and a predominantly white ethnicity which is disproportionately comprised of recent immigrants (i.e. blacks and Latinos) sometimes resemble each other. 

Why is this: Skin color? Location in the economy? Culture, family practices, and values?  The divergent trajectories of Latinos and Asian Americans are not due to historical differences in the scope of white prejudice and discrimination; whites treated Asians at least as badly as Latinos for most of the nineteenth and twentieth centuries.  We know too little about the views of whites about people who are neither white nor black, as well as too little about the views of Latinos and Asians themselves, to do much more than suggest the nature, causes, and consequences of interracial and interethnic attitudes.[19]  But as the demography of the United States changes over the next few decades, so will its politics and social practices (and hopefully its research agendas). The possibilities for racial/ethnic coalitions and contestations could well move to the center of the study of racialized politics.

Hughes and Tuch point to the complications inherent in the study and practice of interracial and interethnic politics, since their research indicates that all four racial/ethnic groups have different views of all four racial/ethnic groups.[20]  They find that groups differ somewhat in their attributions for the poverty of other groups, and in the effect of those attributions on their preferences for racial policy.  They also find, intriguingly, that “minority resentment…emerges… as an important determinant of racial policy attitudes” among minority groups as well as among whites (Hughes and Tuch, this volume, draft results on pp.21-22, quotation on p. 26).  If this finding holds up, it calls into question theories based on the social dominance of whites or on the unacknowledged virtues of whiteness.

A further hint of the complexities to come in multiracial and multiethnic interactions appears in one of the only other national surveys with substantial numbers of respondents from all four racial/ethnic groups. In 1994 members of the four groups were asked which of the other three they felt most and least in common with.  As table 1 shows, the responses were almost perfectly unstable:



Table 1:  Racial and Ethnic Groups’ Affinity for One Another


“Of these groups, if you had to say, which one do you feel you have the most in common with/ least in common with?”   (% choosing each group)





Feel most in common with:








Asian American








African Americans
























Error! Bookmark not defined.







Feel least in common with:





Asian American







African Americans






























Source: National Conference 1994: 32




Whites feel most in common with blacks, who feel little in common with whites.  Blacks feel most in common with Latinos, who feel least in common with them. Latinos feel most in common with whites, who do not feel much in common with them.  Asians feel most in common with whites, who feel least in common with them.  Each group is chasing another which is running from it. 

            If these results obtain in a broad array of political and policy views, American politics in the twenty-first century are going to be interesting, to say the least. Most of our standard theories about the nature of racial prejudice and the direction of racial discrimination may become bouleversé.


The First Cause:  We come finally to the core dispute between Sniderman and his co-authors on the one hand, and most contributors to Racialized Politics on the other.  The former claim that “the fundamental lines of cleavage on issues of race… are not peculiar to issues of race.  They belong, rather, to a larger pattern of division, defining the deeper-lying structure of American politics since the New Deal, centering on the clash of competing conceptions of the proper responsibilities of government and the appropriate obligations of citizens” (draft, p. 4).[21]  Sniderman and his various co-authors begin from the reasonable premise that one must take conservatives at their word as much as, although no more than, one must take liberals, whites, African Americans, or anyone else  at their word.  Thus when conservatives claim that their opposition to government programs to aid African Americans stems from their general opposition to governmental intervention in the society and economy, not from hostility to African Americans or a desire to keep them in their place, Sniderman and his colleagues believe those claims unless and until they can demonstrate their falsity.

            Kinder and Mendelberg, Sears et al., Sidanius et al., and Bobo (to name only the most explicit members of the opposite cluster) start from the opposite premise, and argue that the results of their research support that premise.  As Sidanius et al. (draft p. 42, emphasis in original) put it, “group identity and social dominance values are among the most important factors driving the racial policy attitudes of White Americans…. ‘Race-neutral’ and ‘principled’ belief systems become effective legitimizing foils for the masking of opposition to redistributive policy in terms which appear to be more morally and intellectually defensible.” Bobo rejects the claim that principled conservative opposition to affirmative action is merely rationalization for racism.  But he insists that “[racially-based group] interests are a necessary element of our thinking about the politics of affirmative action….  It has been easy to overplay the argument from principles. Those on the right who wish to don the armor of moral innocence in their war against affirmative action are ready to accept this view” (draft pp. 26, 31). The scholars in this cluster, to put it mildly, do not accept at face value conservative justifications of opposition to government programs to aid blacks, because they do not think it possible for an American to have a political ideology that is neutral on the question of race.

            Here we have the core of all of the other disagreements, and the explanation for much of the intensity that has permeated this debate for years. And here we have the ultimate question of lumping and splitting: can Americans split political ideologies from their racial history, presumptions, and practices, or not?  To put the debate in the terms that I employed over a decade ago (Hochschild 1984), are racial views anomalous to Americans’ political values and ideologies, such that they can be separated from them and changed without doing violence to those values and ideologies?  Or are they symbiotic with Americans’ core values and ideologies, such that the latter are largely created and driven by racial affect, racial position, or racial commitments? 

            Here I will remain perched firmly on the top of the fence, for two reasons.  First, I am less certain that there is a single correct answer than I used to be.  Until the 1960s, the model of symbiosis seemed incontrovertably correct. But the evidence suggests that something deep and broad has happened to whites’ racial attitudes over the past three decades.  Many, perhaps even most, white Americans now believe in the legitimacy of basic racial equality in rights and opportunities.  One must query how many whites hold that belief, what costs -- if any -- they are prepared to pay to put the belief into practice, what contradictory beliefs they hold, and so on. But I see little reason to predict that citizens of the United States would allow our nation to return to anything like a Jim Crow system of segregation and degradation.

            Thus far, most of the authors in Racialized Politics agree.[22]  And if that point is granted, it follows that the basic values of egalitarianism (defined as equal opportunity and equality of core rights), individualism (defined as self-ownership and autonomy of all persons), and liberalism (defined as the legitimacy of governmental intervention to protect core rights, provide equal opportunity, and ensure autonomy) are now substantially independent of racial affect and group commitments – even though they once were not (E. Morgan 1975; Smith 1997; P. Morgan 1998). That conclusion implies nothing about the best policies to sustain or enhance the practice of these values. Americans will surely continue to debate over those questions, but that is a different and somewhat more superficial debate.

            However, the relationship between racial hierarchy and American values has not fully shifted from symbiosis to anomaly, and I do not believe that it ever will. Compelling evidence, ranging from psychological experiments to audits of housing and job opportunities to individual testimony to – yes, even – survey research shows that many white (and perhaps nonwhite) Americans harbor hostility to or seek dominance over black Americans. The point holds for liberals as well as for conservatives, although the ways in which racial hierarchy entwines with ostensibly non-racial values may be different across the two ideologies. What Charles Silberman (1964: XX) wrote three and a half decades ago remains partly true:

The tragedy of race relations in the United States is that there is no American dilemma. White Americans are not torn and tortured by the conflict between their devotion to the American creed and their actual behavior. They are upset by the current state of race relations, to be sure. But what troubles them is not that justice is being denied but that their peace is being shattered and their business interrupted.


Thus one reason that I will not take sides on the dispute over First Causes is that, in my view, the best analysis of racial attitudes should have a “both/and” rather than an “either/or” framing.  I think, in short, that the debate among proponents of principled conservatism, symbolic racism, and social structural approaches has gone as far as it can for the present.

            The other reason for my agnosticism on the First Cause is that I would like to shift our attention away from it, regardless of what position one holds. The contributors to Racialized Politics implicitly if not explicitly agree that “the most complicated aspect of race relations in America today concerns attitude” (Kinder and Sanders 1996: 6). I disagree.  In my view, structural features of American racial politics are at least as important as attitudes. How people behave and how their behavior is constrained or encouraged by phenomena outside their or anyone else’s volition matter as much as, or arguably more than, what people feel or want in determining our (or any nation’s) racial future.  

This claim takes me back to the opening paragraph of this commentary, in which I proposed two types of explanation for my (and others’?) inability or unwillingness to reject two of the three basic approaches articulated in Racialized Politics.  I turn to the second explanation now.


An Explanation of Racial Politics Outside the Survey Research Paradigm

Marylee Taylor’s contribution to Racialized Politics comes closest to my assertion that the structure of constraints and opportunities has more impact on racial politics than do attitudes about race and racial policy.  She examines the effects of the proportion of blacks in their community on white respondents’ racial attitudes, and generates impressive results.  Traditional prejudice, opinions about race policy, racial resentment, beliefs underlying policy preferences, even some views that are ostensibly not related to race -- all are affected by the proportion of blacks living in the communities of white respondents.  “Altogether, examining the sensitivity of white public opinion to the local racial context seems a promising means of detecting the role of racial sentiment in shaping public opinion” (draft p. 23).  Whites may not know the racial composition of their communities – Taylor gives us no evidence on this point – but it deeply influences their views, for the worst from the perspective of African Americans.

            This line of argument begins to move the research agenda in a useful direction, toward a focus on the ways in which structures, institutions, circumstances, and practices differentially affect the life chances of people of different races (and ethnicities?) without anyone necessarily intending it or even recognizing it. Following this path would lead us away from survey research  (as well as focus groups and intensive interviews) and toward analyses of such things as the economic impact of a given level of education, the political effect of electoral laws and metropolitan boundaries, the impact on health and longevity of living in a particular location, the social consequences of demographic dynamics, the trajectory of cultural influences across groups, and so on.  There are, of course, huge literatures on these and other subjects that I cannot begin to review here.  I will conclude instead with something of an obiter dictum about what they tell us.

            Race matters.  The contributors to Racialized Politics and almost all Americans agree.  But race matters in complicated ways, about which there is plenty of disagreement.  In my view, four features stand out from the welter of conflicting evidence and assertions:[23]

  • African Americans are and will continue to be at a disadvantage compared with all other Americans[24] on most dimensions that affect a person’s life chances – the range of plausible marriage partners, the prospect of living in any community that one chooses, the opportunity to reach the top of one’s profession, the chance of attaining a high political office, the simple option of disappearing in a crowd when one wants to.  Most Asian Americans are close to becoming “honorary whites,” in Andrew Hacker’s inimitable phrase, and within a few decades the trajectory of Latinos will come to resemble the trajectory of white ethnic immigrants after the turn of the century.
  • Class matters deeply among African Americans, but not in the same way that it matters for people of other races and ethnicities.  To begin with, there is greater economic and social disparity between wealthy and poor blacks than between wealthy and poor whites, and that disparity is growing.  A nontrivial proportion of poor blacks live in circumstances that are worse than those of all other groups, and I see no evidence that most Americans are willing to do what would be necessary to improve those circumstances.  Nevertheless, it is the relatively well-off African Americans who are most alienated from American society, and most likely to despair about the eventual attainment of racial equality and justice.  That is historically unprecedented for a recently upwardly-mobile group, and it augurs very badly for American politics and society unless something changes soon.
  • Regardless of their racially-oriented intentions, nonblacks behave in ways that reinforce disparities between blacks -- especially poor blacks -- and all other groups.   Conservatives are content to leave too many outcomes to be determined by markets, which are as likely to reinforce as to offset initial economic inequalities; they fail to see how social structures and practices benefit those who were born lucky and harm those who were not.  Liberals allow too many decisions to be made by political forces, which are as likely to reinforce as to offset initial political inequalities; they fail to see ways in which their own behavior benefits people like themselves at the expense of people unlike themselves.  Possessors of both viewpoints are more inclined to blame the other side than to make painful ideological and behavioral changes in order to ameliorate America’s racial and class problems.
  • In short, inequality between black Americans and all other Americans is embedded in our history, economic processes, political institutions, unexamined assumptions, and norms.  Pejorative racial attitudes, the desire for dominance, or opposition to liberal racial policy proposals – none are necessary to maintain this inequality, and probably none would not suffice to keep it in place if structures and processes somehow shifted to provide disproportionate benefits in the opposite direction.

I will conclude this contentious list of assertions with one that moves in the opposite direction, and that might help to keep all of us working as hard as the authors in Racialized Politics do to foster racial justice:

  • Real movement toward racial equality has occurred in all of these features of the American polity, as well as in individuals’ attitudes.  That movement was enhanced if not created by extraordinary efforts of identifiable individuals, ranging from Civil War soldiers to Supreme Court justices to civil rights activists to corporate leaders seeking profit from diversity.  Racial inequality is deeply embedded in Americans’ practices, and possibly in their views and values, but it is not immovably so.  We did move, for the better, at various points in our history and we can do so again.  Attitudes will matter for that effort, although they will not be all that will matter.






Conover, Pamela and Donald Searing (1998). “Political Discussion and The Politics of Identity," paper presented at the annual meeting of the Midwest Political Science Association, Chicago IL


Conover, Pamela, Donald Searing, and Ivor Crewe (1998) "Tolerance and Community in the United States and Great Britain." paper presented at the annual meeting of the American Political Science Association, Boston MA.


d'Souza, Dinesh (1995). The End of Racism. New York: Free Press.


Fredrickson, George (1995). Black Liberation: A Comparative History of Black Ideologies in the United States and South Africa. New York: Oxford University Press.


Fried, Amy (1997) Muffled Echoes: Oliver North and the Politics of Public Opinion.  New York: Columbia University Press.


Gilroy, Paul (1993). The Black Atlantic: Modernity and Double Consciousness. Cambridge MA: Harvard University Press.


Hibbing, John and Elizabeth Theiss-Morse. (1995). Congress as Public Enemy: Public Attitudes toward Political Institutions. New York: Cambridge University Press.


Hochschild, Jennifer (1981) What's Fair: American Beliefs about Distributive Justice. Cambridge MA: Harvard University Press.


Hochschild, Jennifer (1984).  The New American Dilemma: Liberal Democracy and School Desegregation.  New Haven CT: Yale University Press.


Hochschild, Jennifer (1993). "Disjunction and Ambivalence in Citizens' Political Outlooks," in George Marcus and Russell Hanson, eds. Reconsidering the Democratic Public. University Park, PA: Pennsylvania State University Press: 187-210.


Hochschild, Jennifer (1995). Facing Up to the American Dream: Race, Class, and the Soul of the Nation. Princeton: Princeton University Press.


Hochschild, Jennifer and Reuel Rogers (1998). "Race Relations in a Diversifying Nation.," in James Jackson, ed.,  New Directions: African Americans in a Diversifying Nation. under review.


Kinder, Donald and Lynn Sanders (1996). Divided by Color: Racial Politics and Democratic Ideals. Chicago: University of Chicago Press.


Knight, Kathleen (1998) "In Their Own Words: Citizens' Explanations of Inequality Between the Races," in Jon Hurwitz and Mark Peffley, eds.,Perception and Prejudice: Race and Politics in the United States. New Haven CT: Yale University Press:  202-232.


Krysan, Maria (1995). White Racial Attitudes: Does It Matter How We Ask? Ph. D. Dissertation, Department of Psychology. Ann Arbor, MI, University of Michigan.


Lamont, Michele (1992). Money, Morals, and Manners: The Culture of the French and American Upper-Middle Class. Chicago: University of Chicago Press.


Lane, Robert. (1962) Political Ideologies. New York: Free Press of Glencoe.


Marx, Anthony (1998). Making Race and Nation: A Comparison of South Africa, the United States, and Brazil. New York: Cambridge University Press.


Morgan, Edmund (1975). American Slavery, American Freedom: The Ordeal of Colonial Virginia. New York: Norton.


Morgan, Philip (1998) Slave Counterpoint:Black Culture in the Eighteenth Century Chesapeake and Lowcountry. Chapel Hill NC: University of North Carolina Press.


Motley, Constance Baker (1998) Equal Justice Under Law.  New York: Farrar, Straus & Giroux.


National Conference of Christians and Jews (1994). Taking America's Pulse: The Full Report of the National Conference Survey on Inter-Group Relations. New York: National Conference of Christians and Jews.


Parikh, Sunita (1997) The Politics of Preference: Democratic Institutions and the Politics of Preference in the United States and India. Ann Arbor MI: University of Michigan Press.


Rae, Douglas, Douglas Yates, et al. (1981). Equalities. Cambridge MA: Harvard University Press.


Reskin, Barbara (1998). Affirmative Action. Washington, DC: American Sociological Association.


Sigel, Roberta (1996). Ambition and Accommodation: How Women View Gender Relations. Chicago: University of Chicago Press.


Smelser, Neil (1998).  "The Rational and the Ambivalent in the Social Sciences."  American Sociological Review. 63 (1): 1-16.


Smith, Rogers (1997). Civic Ideals: Conflicting Visions of Citizenship in U.S. History. New Haven CT: Yale University Press.


Sniderman, Paul and Edward Carmines (1997). Reaching Beyond Race. Cambridge MA: Harvard University Press.


Sniderman, Paul and Thomas Piazza (1993). The Scar of Race. Cambridge, MA: Harvard University Press.


Steeh, Charlotte and Maria Krysan (1996). “The Polls -- Trends.  Affirmative Action and the Public, 1970-1995.” Public Opinion Quarterly 60 (1): 128-58.


Zaller, John (1992). The Nature and Origins of Mass Opinion. New York: Cambridge University Press.




Additional Items for Bibliography, “Lumpers and Splitters…”

Jennifer L. Hochschild

April 2, 1999



Hacker, A. (1992). Two Nations: Black and White, Separate, Hostile, Unequal, Scribner's.


Hochschild, J. (1991). “The Politics of the Estranged Poor.” Ethics 101(3): 560-78.


Hochschild, J. and R. Rogers (1999, forthcoming). "Race Relations in a Diversifying Nation." New Directions: African Americans in a Diversifying Nation. J. Jackson, ed.  Washington D.C., National Planning Association.


Jackman, M. and M. Muha (1984). “Education and Intergroup Attitudes.” American Sociological Review 49(6): 751-769.


Kelley, S. (1983). Interpreting Elections. Princeton NJ, Princeton University Press.


Mills, C. (1997). The Racial Contract. Ithaca NY, Cornell University Press.


Silberman, C. (1964). Crisis in Black and White. New York, Random House.



[1] Sniderman and his colleagues are splitters in the realm of policy choice as well as in the psychological realm: “there is not one problem of race, but a number -- among them, the need to assure equal treatment under the law; to provide assistance to those who are badly-off; to combat discrimination, which in itself can take different forms as discrimination itself is differently defined…” (draft, p. 3).

[2] Paralleling my observation in note #1, Sears and Henry are lumpers in the realm of policy choices as well as in the psychological realm: “it’s [that is, anti-black affect’s] effects are very similar across quite different domains of racial policies, suggesting that the common racial component rather than the individual political contents of these policies most captures whites’ attention.” (draft, p. 39).

 Sidanius and his colleagues are also lumpers rather than splitters, although along somewhat different dimensions than the other two central models.  For example, “opposition to policies designed to aid subordinates is driven by very similar forces and dynamics across a number of different social systems…. Group-based anti-egalitarianism and SDO [social dominance orientation] are critically important sources of opposition to group-relevant social policy attitudes across a wide variety of other cultures as well [as the United States].” (draft, p. 48)

[3] “Racism is an ideology of intellectual or moral superiority based upon the biological characteristics of race.” [d'Souza, 1995 #1044: 27]

[4] “Those who resist [affirmative action programs] deny that they are racists, but the truth is that their real motivation is racism, a belief in the inherent inferiority of African-Americans and people of mixed racial backgrounds” (Motley 1998: XX)

[5] Kinder and Sanders use the same NES questions with the words “preference” and “quotas,” despite their discussion of Laura Stoker’s finding that support for affirmative action increases when the question more carefully tracks Supreme Court rulings (1996: 174-182, 194-195). Sears et al. also use the NES questions, so their chapter is subject to the same critique.

[6] In the chapter in Racialized Politics, Sniderman et al. conduct a detailed analysis of whites’ attitudes toward “black leaders asking for affirmative action.”  Although their interest centers on whether that phenomenon makes whites angry, they make no effort to discover what angry (or nonangry) whites mean by “affirmative action.”

[7] The problem in Sniderman and Carmines (1997) is more subtle than the problem in Sniderman and  Piazza (1993).  In the later book, the authors elegantly demonstrate (pp. 22-27) just what Bobo points out – that a majority of whites support “extra effort” although a majority oppose “preferential treatment.”  They nevertheless equate affirmative action with quotas and preferences throughout the rest of the book, rather than with special efforts.

[8] Kinder and Mendelberg discuss Mary Jackman’s rather different understanding of individualism in their chapter r for this volume, but they focus more on relating it to their extant theory than to use it or other meanings as a fresh start in developing new theories.

[9] Rae et al.(1981) sort out the many other meanings of equality, and the ways in which some can directly offset others.

[10] (Draft, p. 17).  “Other”might mean non-American; how should we locate them on a social dominance scale?

[11] Pettigrew treats individual racial or ethnic prejudice as a dependent variable, whereas most of the other authors in Racialized Politics use it as an independent variable.  I will not discuss that point, since it is simply a choice of research agenda rather than part of an on-going dispute which these comments seek to understand and negotiate.

[12] In this case, I am a lumper (I argue for combining theories as much as possible) rather than a splitter (who would argue for separating out the one best theory from the other, less effective, alternatives).

[13] Alternatively, one could argue that the three basic approaches in Racialized Politics are “key conceptualizations” that “organize the field,” but that we do not yet have a “systematic” or “definitive” theory of racial interactions that could integrate all of the relevant elements (Pettigrew,this volume, draft pp. 1-2).

[14] Surveyors, as well as survey analysts, vary greatly and unsystematically in how they deal with people who are neither white Americans nor African Americans.  Some surveys and some theories distinguish between whites and minorities (which typically include Latinos, Asians, and “others”).  In other surveys and theoretical frameworks, the focus lies on whites and blacks (in which case Latinos, Asians, and others are excluded from the survey or from the analysis altogether). In still other surveys and frameworks, the distinction is between blacks and nonblacks (in which case Latinos etc. are included in the “white” population).  In this chapter, I am lumping all of these possibilities together into the category of bifocal analysis, to be distinguished from a multifocal analysis as described in the text.

[15] It would be worthwhile for Kinder to reconcile this finding with that in Kinder and Sanders (Kinder and Sanders 1996: 272) that “our most striking result is the long reach of racial resentment into diverse aspects of American opinion.  We found racial resentment to be implicated in whites’ views not just on affirmative action or school desegregation, but on welfare, capital punishment, urban unrest, family leave, sexual harassment, gay rights, immigration, spending on defense, and more.”

[16] In an alternative phrasing, “the fundamental lines of cleavage on issues of race, so far as they are defined by the party system, are not peculiar to issues of race. They belong, rather, to a larger pattern of division, defining the deeper-lying structure of American politics since the New Deal, centering on the clash of competing conceptions of the proper responsibilities of government and the appropriate obligations of citizens”  (this draft, p. 4).

[17] The terms “Asian,” “Hispanic,” “Latino” or other related words do not appear in the index of either the 1993 or the 1997 book, and neither book discusses racial or ethnic groups other than blacks and whites.  The same obtains for Kinder and Sanders (1996).

[18]  She differs from those just discussed in a crucial way that I will address below, but for now I am focusing on the resemblance.

[19] There is no national survey of Asians; there are only two national surveys of Latinos, the most recent of which is a decade old.

[20] The lack of national surveys in which all four groups are surveyed symmetrically about all four groups is intensely frustrating.  Even the 1995 Washington Post survey, which Hughes and Tuch use and which is the best source of national data on the attitudes of all four groups, missed a major opportunity by not asking the four groups how they explained white poverty.  Thus the four-way comparison that we need in order to understand how “poverty influence[s] racial policy attitudes” is unavailable. 

This is an instance of a larger complaint I have about survey research – surveyors ask endless questions about the poor but few about the rich, many questions about women but very few about men, a multitude of questions about blacks (and increasingly about Latinos and Asians) but almost none about whites.  Postmodernists write about The Gaze; surveyors seldom turn it onto people like themselves.

[21] Or, more epigrammatically “the contemporary debate over racial policy is driven primarily by conflict over what government should try to do, and only secondarily over what it should try to do for blacks (Sniderman and Carmines 1997: 4, their italics).

[22]  With perhaps the crucial exception of the two senior black contributors.

[23] Much of the evidence for and analyses underlying the following claims is in Hochschild 1995.

[24] Excepting Native Americans