Intergroup Attitudes As the Property of Groups

The third distinctive feature of my empirical framework lies in the shift from the individual to the group as the primary focus of attention. Previous research on intergroup attitudes and group consciousness has overwhelmingly focused on individual-level variance in attitudes in the endeavor to answer questions about the causes of those attitudes. Thus, the enduring empirical research questions have been in the following vein: Does an increase in years of formal education make whites less prejudiced toward blacks? Does an increase in personal contact with blacks break down whites' racial prejudice? Are certain personality types (especially authoritarians) more likely to be prejudiced or intolerant? What individual life experiences account for some women having less "traditional" sex-role attitudes than others? What factors in an individual's experience in the workplace (e.g., authority relations, unionization) influence the articulation of class consciousness?

Without denying the validity of variance among individuals, I wish to explore a different avenue of explanation. In this study, I shift the primary empirical burden to the aggregate level of the intergroup relationship taken as a whole. My strategy is to identify the prevailing attitudinal properties of each intergroup relationship and to focus attention on the

variance that exists among different kinds of intergroup relationships. My purpose is to uncover and explore sources of variance that cannot be identified within the empirical framework of past research.

There is a necessary relationship between a specific methodology and the substantive questions that come into focus, and the present case is no exception. My shift from the individual to the group as the primary focus of attention invokes both a different methodological strategy and a different substantive emphasis from that common to past research. Individual variance exists within groups both in the intergroup attitudes that are espoused and in pertinent experiential factors. Past research sought to exploit that variance. Such an approach, of course, assumes that the variation that exists within groups captures a broad enough range to be empirically useful. At the same time, that approach places an implicit emphasis on the individual actor as the source of intergroup attitudes. This has encouraged a relatively atomized conception of the individual actor: it is assumed that something should be experienced directly and personally to enter the salient realm of human cognizance.

That assumption is evidenced in standard analytic strategies. It is taken as a given that if a proposition is valid, it will hold true at the individual level. For example, the hypothesis that intergroup contact affects intergroup attitudes has been tested by comparing those members of a group who have had personal intergroup contact with those who have not (rather than comparing groups from different intergroup relationships where the prevailing experiences are different). Similarly, it is routinely anticipated that if authority relations affect workers' level of consciousness, this should be demonstrated empirically by an observable difference in attitudes between workers whose personal experience has been with different kinds of authority relations (rather than examining how the prevailing mode of authority relations may condition the consciousness of workers as a whole). Analysts have sought to test theories by seeing whether they have explanatory power at the individual level.

This has especially been the case in the analysis of racial prejudice, where the phenomenon itself has been conceived as an expression of individual experiential and psychological factors. Thus, when Allport asked, "Prejudice: Is it Societal or Personal?" (1962), he was not posing a choice between the group and the individual as the unit of analysis, but between the individual acting according to the dictates of his personality (for example, prejudice as a function of "authoritarianism") and the individual responding to factors that derive from his social experiences (for example, prejudice as a function of formal educational attainment). In a somewhat similar vein, Tajfel's lengthy discussion of "Individuals and Groups in Social Psychology" (1981) essentially poses the question of the extent to which individual behavior is driven by indige-

nous biological-perceptual proclivities or is a response to factors acting on the individual from the social environment in which he lives.

In an early commentary on the prejudice literature, Blumer argued that this emphasis on the individual level of experience was misplaced:

The dominant group is not concerned with the subordinate group as such but it is deeply concerned with its position vis-a-vis the subordinate group. . . . The sense of group position is the very heart of the relation of the dominant to the subordinate group. . . . To seek . . . to understand [race prejudice] or to handle it in the arena of individual feeling and of individual experience seems to me to be clearly misdirected. (1958, 4, 7)

Blumer's comments had almost no impact on the research agenda on prejudice. That agenda continued to be dominated by theories that identified individual differences in personality or experience as the root of the problem. Interestingly, even those empirical analyses that have been most influenced by Blumer's work (analyses based on realistic group-conflict theory) have still been drawn to individual-level variance as the explanatory wedge (see, e.g., Wellman 1977; Smith 1981; Bobo 1983; Giles and Evans 1986).

Blumer's notion of "the sense of group position" was perhaps too obliquely defined to have utility as a sharp analytical tool. He stopped short of seeing dominant groups as being driven by tangible interests; his conception of prejudice emphasized the same features as did mainstream conceptions, except that they were transposed to an aggregated level. Hostility toward and derogation of subordinate groups were attributed to a collective cultural process that dealt with subordinates as a symbolic abstraction. Blumer's "sense of group position" was very much a "sense" rather than a tangible "position," which made it an aggregate-level analog of individual-level prejudice. Because it was not linked to specific aspects of the structure of the relationship between groups, the "sense of group position" did not carry a clear enough set of empirical referents to suggest a particular line of empirical inquiry. But Blumer's idea that the problem of prejudice was located in a collective group process did run directly against the prevailing social-reformist current of research on prejudice. That current mandated the identification of specific individual-level factors that might be manipulated to reduce prejudice.

Research on gender and class attitudes has also placed the overwhelming explanatory burden on individual-level variation, even though the phenomenon of interest is not seen as so exclusively within the psychological realm. The literature on class consciousness, in particular, has been shaped by theories that address the macrostructure of society. Yet individual-level motivational factors are still regarded as a critical element, and those factors are seen as primarily the product of direct individual

experiences (e.g., Lockwood 1966). Indeed, the common emphasis among students of class, race, and gender attitudes on hostility as a vital ingredient of expropriative intergroup relations is consistent with the broader analytic assumption that the locus of energy is in the individual. If one casts the attitudes of the group as the simple sum of its members' individually generated attitudes, personal feelings of hostility make an intuitively plausible motivational basis for the existence of discriminatory dispositions.

The assumption that the critical energy lies at the individual level is so ingrained in our way of thinking about intergroup attitudes that it is even reflected in the basic methodology of questionnaire-item construction. As Weissberg (1976) has pointed out, a requisite characteristic of good questionnaire items is that they elicit variance in responses, and it is a cardinal rule to avoid items that do not generate reasonable variance. Although this rule is dictated by the pragmatic need for individual-level variance if that is to be the fuel for analyses, it clearly has conditioned the way we define empirical problems. Issues that arouse consensual responses from all the members of a group reveal something about the shared experience of the group. Similarly, issues that elicit consensus from all participants in a relationship tell us about the prevailing values that encase the relationship and that define the limits of the political agenda (Weissberg 1976). Such issues may be especially germane to our understanding of the central currents in an intergroup relationship, but in explanations that hinge on individual-level variance, they are dismissed as "constants" with no utility.

I do not mean to deny that individual variance exists on salient issues or that differences in individual experiences bear on the expression of intergroup attitudes. However, exclusive attention to such individual variation, especially in the context of a single group in a single relationship (for example, whites' racial attitudes or women's gender attitudes), draws the explanatory endeavor away from questions about the overall structure of the relationship. The latter may impinge deeply on the participants in the relationship, constraining the amount of variance to be observed within a single group or relationship.

Consider the question of how the degree of personal contact between groups affects intergroup attitudes. This question has been studied primarily via individual variance among whites in their personal contact with and attitudes toward blacks, and yet there is only limited variance in whites' contact with blacks. Most whites have little or no personal contact with blacks and the intimate end of the continuum is marked by a small minority of whites who can name one friend who is black (or occasionally two). Contrast this with the gender relationship, where the overwhelming norm for intergroup intimacy far exceeds the extreme

end of the continuum for race and where the physical separation from subordinates that most whites experience is a rare event for men. Either relationship by itself captures only a portion of the empirical range of human experience that exists across relationships. The variance between the two relationships expands the observable range dramatically and thus offers the opportunity to measure the effects of intergroup contact over a broader spectrum.

The shift to variance between intergroup relationships as the explanatory source, whether or not it increases the observable range of experience, carries with it a consequential shift in the kinds of substantive questions that come into focus. It draws attention to the structural conditions that encase an intergroup relationship and it underscores the point that individual actors are not free agents but are caught in an aggregate relationship. Unless we assume that the individual is socially atomized, her personal experiences constitute only one source of information that is evaluated against the backdrop of her manifold observations of the aggregated experiences (both historical and contemporaneous) of the group as a whole. Even if we are willing to assume that the individual is motivated entirely by selfish concerns (which I am not willing to do), it would still be hasty to assume that her inferences about how best to serve those concerns are based solely or even primarily on her direct experiences. As I argued in the prologue, because the individual is situated within social relations, she cannot help but draw inferences about her own fate from her observations of the experiences of others, especially if she has reason to believe that her fate is linked (either positively or negatively) to those others. Individuals do not have to wait until they are personally knocked down by a hurricane to understand that hurricanes are dangerous.

This is not to say that all members of a group are expected to think the same way. The structural constraints imposed by the organization of an intergroup relationship are not felt uniformly by all members of a group. People are not impervious to their direct experiences. And information is not dispersed evenly to all members of a group. There are channels of communication that systematically influence the information to which individuals are exposed. For example, educational institutions influence people's cognitive style and the kinds of ideas and information available to them: dominant-group members with an advanced formal education are thus likely to espouse a more elaborate and sophisticated ideology to defend their group's interests (Jackman and Muha 1984). Similarly, direct personal contact with members of a subordinate group alters the information available to individual dominant-group members and has been shown to influence some aspects of their intergroup attitudes (Deutsch and Collins 1951; Wilner, Walkley and Cook

1955; Jackman and Crane 1986). Some information also trickles through the group along irregular and unpredictable paths depending on the particular constellation of personal characteristics and experiences that each individual has from childhood through his adult life. And as these individuals meet and interact with one another, some idiosyncratic networks of communication are created that could be either durable or ephemeral. thus, it should not be surprising that analyses using individual-level data have found that no single explanatory model can account for more than a small portion of the variance among group members in their intergroup attitudes.

The intergroup attitudes that people espouse are not their personal property, devised from their own experience independently, but the property of the group to which they belong. The group's collective experience provides the informational context within which individuals interpret their personal experiences. For although individual characteristics and experiences influence personal gravitation to one kind of attitude or another from those that are arrayed in the collective culture, the array is not infinite but is instead constrained by the limits of the collective experience. As ideas appear, they gain or lose currency according to their efficacy in meeting the political needs of the moment. They may spread rapidly, fizzle out, or gradually become reshaped over time. At any single point in history, the ideological baggage of the group as a whole reflects all these currents, as new ideas are heaped on top of old ones that then become discarded (rapidly or slowly) as they lose their political utility.

As we examine the pattern of attitudes that characterizes each group, we learn about the collective condition of the group as it faces another. We can then ask whether the position of a group in a relationship (dominant or subordinate) produces any systematic tendencies in attitudinal dispositions, or whether completely different rules seem to govern the exchange between groups from one intergroup relationship to another. We can observe whether unequal groups face off in competing, internally unified camps, whether they join together in a common consensus, or whether they are engaged in muted conflict that displays internal divisions within each camp or shifts from intergroup consensus to conflict from one kind of issue to another. By examining such questions across different cases of inequality that manifest widely variant structures, we can formulate ideas about the generality of the attitudinal processes that accompany inequality and their sensitivity to alternative structural pressures.