Final Project

The Social Identity Theory of Intergroup Behavior • 277

READING 16

•

again, it is probable that pure forms of this ex- treme are found only infrequently in real social situations. Examples that might normally tend to be near the interpersonal extreme would be the relations between wife and husband or between old friends. Examples that would normally ap- proach the intergroup extreme are the behavior of soldiers from opposing armies during a battle, or the behavior at a negotiating table of members rep- resenting two parties in an intense intergroup con- flict.

Some of the theoretical issues concerning this continuum are discussed by Turner (1982, 1984), Brown & Turner (1981), and Stephenson (1981); the main empirical questions concern the condi- tions that determine the adoption of forms of so- cial behavior nearing one or the other extreme. The first-and obvious-answer concerns intergroup conflict. It can be assumed, in accordance with our common experience, that the more intense is an intergroup conflict, the more likely it is that the individuals who are members of the opposite groups will behave toward each other as a func- tion of their respective group memberships, rather than in terms of their individual characteristics or interindividual relationships. This was precisely why Sherif (1967, for example) was able to abol- ish so easily the interindividual friendships formed in the preliminary stages of some of his field stud- ies when, subsequently, the individuals who had become friends were assigned to opposing groups.

An institutionalized or explicit conflict of ob- jective interests between groups, however, does not provide a fully adequate basis, either theoretically or empirically, to account for many situations in which the social behavior of individuals belong- ing to distinct groups can be observed to approach the "group" extreme of our continuum. The con- flict in Sherif's studies was "institutionalized" in that it was officially arranged by the holiday camp authorities; it was "explicit" in that it dominated the life of the groups; and it was "objective" in the sense that, given the terms of the competition, one of the groups had to be the winner and the other group the loser. And yet, there is evidence from Sherif's own studies and from other research that the institutionalization, explicitness, and objectiv- ity of an intergroup conflict are not necessary conditions for behavior in terms of the "group" extreme, although they will often prove to be suf-

1964). Thus, real conflicts of group interests not only create a~tago?istic. inter?roup relati?~s but also heighten IdentificatIOn WIth, and pOSItIVe at- tachment to, the in-group.

This identification with the in-group, however, has been given relatively little prominence in RCT as a theoretical problem in its own right. The de- velopment of in-group identifications is seen in RCT almost as an epiphenomenon of intergroup conflict. As treated by RCT, these identifications are associated with certain patterns of intergroup relations, but the theory does not focus either upon the processes underlying the development and maintenance of group identity nor upon the possi- bly autonomous effects upon the in-group and in- tergroup behavior of these "subjective" aspects of group membership. It is our conte?tion th~t the relative neglect of these processes m RCT IS re- sponsible for some inconsistencies between the empirical data and the theory in its "classical" form. In this sense, the theoretical orientation to be outlined here is intended not to replace RCT, but to supplement it in some respects that seem to us essential for an adequate social psychology of intergroup conflict-particularly as the under- standing of the psychological aspects of social change cannot be achieved without an appropriate analysis of the social psychology of social conflict.

The Social Context of Intergroup Behavior

Our point of departure for the discussion to fol- low will be an a priori distinction between two extremes of social behavior, corresponding to what we shall call interpersonal versus intergroup be- havior. At one extreme (which most probably is found in its pure form only rarely in real life) is the interaction between two or more individuals that is fully determined by their interpersonal re- lationships and individual characteristics, and not at all affected by various social groups or catego- ries to which they respectively belong. The other extreme consists of interactions between two or more individuals (or groups of individuals) that are fully determined by their respective member- ships in various social groups or categories, and not at all affected by the interindividual personal relationships between the people involved. Here

The alternative to these approaches has been represented by the work of Muzafer Sherif and his associates and has been referred to by D. 1. Campbell (1965) as the "realistic group conflict theory" (RCT). Its point of departure for the ex- planation of intergroup behavior is in what Sherif (1967) has called the functional relations between social groups. Its central hypothesis-"real con- flic't of group interests causes intergroup con- flict"-is deceptively simple, intuitively convinc- ing, and has received strong empirical support (including Avigdor, 1953; Bass & Dunteman, 1963; Blake & Mouton, 1961, 1962; Diab, 1970; Harvey, 1956; Johnson, 1967; Sherifet al., 1961; Sherif & Sherif, 1953).

RCT was pioneered in social psychology by the Sherifs, who provided both an etiology of inter- group hostility and a theory of competition as re- alistic and instrumental in character, motivated by rewards which, in principle, are extrinsic to the intergroup situation (see Deutsch, 1949; Julian, 1968). Opposed group interests in obtaining scarce resources promote competition, and positively in- terdependent (superordinate) goals facilitate co- operation. Conflicting interests develop, through competition, into overt social conflict. It appears, too, that intergroup competition enhances intra- group morale, cohesiveness, and cooperation (Fiedler, 1967; Kalin & Marlowe, 1968; Vinacke,

The Social Identity Theory of Intergroup Behavior

Henri Tajfel • Formerly of the University of Bristol, England John C. Turner. Macquarie University, Australia

Introduction

The aim of this chapter is to present an outlineof a theory of intergroup conflict and some preliminary data relating to the theory. First, how- ever, this approach to intergroup behavior and in- tergroup conflict must be set in context, in rela- tion to other approaches to the same problem.

Much of the work on the social psychology of intergroup relations has focused on patterns of in- dividual prejudices and discrimination and on the motivational sequences of interpersonal interac- tion. Outstanding examples of these approaches can be found, respectively, in the theory of authori- tarian personality (Adorno et al., 1950) and in the various versions and modifications of the theory of frustration, aggression, and displacement (such as Berkowitz, 1962, 1969, 1974). The common denominator of most of this work has been the stress on the intraindividual or interpersonal psy- chological processes leading to prejudiced atti- tudes or discriminatory behavior. The complex interweaving of individual or interpersonal behav- ior with the contextual social processes of inter- group conflict and their psychological effects has not been in the focus of the social psychologist's preoccupations (see Tajfel, 1981, pp. 13-56, and Turner & Giles, 1981, for more detailed discus- sions).

276

The Social Identity Theory of Intergroup Behavior • 279278 • Political Psychology

ficient conditions. One clear example is provided by our earlier experiments (Tajfel, 1970; Tajfel et al., 1971), which we shall discuss briefly below, in which it was found that intergroup discrimina- tion existed in conditions of minimal in group af- filiation, anonymity of group membership, absence of conflicts of interest, and absence of previous hostility between the groups.

Other social and behavioral continua are asso- ciated with the interpersonal-intergroup con- tinuum. One of them may serve to summarize a quasi-ideological dimension of attitudes, values, and beliefs that may be plausibly hypothesized to play a causal role in relation to it. This dimension will also be characterized by its two extremes, which we shall refer to as "social mobility" and "social change." These terms are not used here in their sociological sense. They refer instead to in- dividuals' belief systems about the nature and the structure of the relations between social groups in their society. The belief system of "social mobil- ity" is based on the general assumption that the society in which the individuals live is a flexible and permeable one, so that if they are not satis- fied, for whatever reason, with the conditions im- posed upon their lives by membership in social groups or social categories to which they belong, it is possible for them (be it through talent, hard work, good luck, or whatever other means) to move individually into another group that suits them better. A good example of this system of beliefs, built into the explicit cultural and ideological tra- ditions of a society, is provided in the following passage from Hirschman (1970):

The traditional American idea of success confirms the hold which exit has had on the national imagi- nation. Success-or, what amounts to the same thing, upward social mobility-has long been conceived in terms of evolutionary individualism. The successful individual who starts out at a low rung of the social ladder, necessarily leaves his own group as he rises; he "passes" into, or is "ac- cepted" by, the next higher group. He takes his immediate family along, but hardly anyone else. (pp. 108-109)

At the other extreme, the belief system of "so- cial change" implies that the nature and structure of the relations between social groups in the soci- ety is characterized by marked stratification, mak- ing it impossible or very difficult for individuals,

as individuals, to.d~vest themsel~es of an unsatis_ factory, u~derpnvtleged,.or stIg~atized grou membership. The economic or socIal realities 0/ society may be such (as, for example, in the cas

a

o.f the millions of unemploy~dduri~g the Depres~ slon of the 1930s) that the ImpOSSibility of "get- ting out" on one's own, as an individual, become an everyday reality that determines many form~ of intergroup social behavior. But even this ex- ample is still relatively extreme. Many social in- tergroup situations that contain, for whatever rea- sons, strong elements of stratification perceived as such may tend to move social behavior away from the pole of interpersonal patterns toward the pole of intergroup patterns. This is as true of groups that are "superior" in a social system as of those that are "inferior" in it. The major characteristic of social behavior related to this belief is that, in the relevant intergroup situations, individuals will not interact as individuals, on the basis of their individ- ual characteristics or interpersonal relationships, but as members of their groups standing in certain defined relationships to members of other groups.

Obviously, one must expect a marked correla- tion between the degree of objective stratification in a social system (however measured) and the social diffusion and intensity of the belief system of "social change." This, however, cannot be a one- to-one relationship for a number of reasons, some of which will be discussed below, although we cannot in this chapter go into the details of the many social-psychological conditions that may determine the transition in certain social groups from an acceptance of stratification to behavior characteristic of the intergroup pole of our first continuum-that is, to the creation of social move- ments aiming to change (or to preserve) the status quo (see Tajfel, 1978a; Giles & 10hnson, 1981, provide a thorough discussion of this issue in the context of seeking to predict the conditions under which ethnic groups will accentuate their distinc- tive languages, dialects, or accents).

It may be interesting, however, to point to the close relationship that exists between an explicit intergroup conflict of interests, on the one hand, and the "social change" system of beliefs on the other. One of the main features of this belief system is the perception by the individuals concerned that it is impossible or extremely difficult to move in- dividually from their own group to another group. This is precisely the situation in an intense con-

flict of inter.ests, in which .it is extremely. d~f!icult for an indi;l~ual to conceive of the pos~lbtllty of "betraying' hIS or her own group by movmg to the opposing group. Although this does happen on occasion, sanctions for such a move are, on the

hole, powerful, and the value systems (at least ~ our cultures) are in fl~grant opposi.tion to it. To use an example from social-psychological research, it seems har.dly possible that one ~f the boys in Sherif's holIday camps would deCide to change sides, even though some of his previously con- tracted friendships overlapped group boundaries.

The intensity of explicit intergroup conflicts of interests is closely related in our cultures to the deoree of opprobrium attached to the notion of "r;negade" or "traitor." This is why the belief sys- tems corresponding to the "social change" extreme of our continuum are associated with intense inter- oroup conflicts. These conflicts can be conceived, ~herefore, as creating a subclass or a subcategory of the subjective intergroup dichotomization char- acteristic of that extreme of the belief continuum. They share the basic feature of the "social change" system of beliefs, in the sense that the multigroup structure is perceived as characterized by the ex- treme difficulty or impossibility of an individual's moving from one group to another.

The continuum of systems of beliefs discussed so far represents one conjecture as to one impor- tant set of subjective conditions that may shift so- cial behavior toward members of out-groups be- tween the poles of "interpersonal" and "intergroup" behavior within particular situations and societies. To conclude this part of our prelimi- nary discussion, we must characterize briefly two further and overlapping continua, which can be considered as encompassing the major conse- quences of social behavior that approaches one or the other end of the interpersonal-intergroup con- tinuum. They both have to do with the variability or uniformity within a group of behavior and atti- tudes concerning the relevant out-groups. The first may be described as follows: The nearer members of a group are to the "social change" extreme of the belief-systems continuum and the intergroup extreme of the behavioral continuum, the more uniformity they will show in their behavior toward members of the relevant out-group; an approach toward the opposite extremes of both these con- tinua will be correspondingly associated with greater in-group variability of behavior toward

members of the out-group. The second statement is closely related to the first: the nearer members of a group are to the "social change" and the "in- tergroup" extremes, the more they will tend to treat members of the out-group as undifferentiated items in a unified social category, rather than in terms of their individual characteristics. The vast literature in social psychology on the functioning of group stereotypes in situations of intense intergroup ten- sions is no more than an example of this general statement.

Thus, this preliminary conceptualization repre- sents an approach to the social psychology of in- tergroup relations that takes into account social realities as weB as their reflection in social behav- ior through the mediation of socially shared sys- tems of beliefs. This convergence occurs at both ends of the sequence just discussed; at the begin- ning, because it can be assumed without much dif- ficulty that the "social change" belief system is likely to reflect either an existing and marked so- cial stratification or an intense intergroup conflict of interests, or both; at the end, because the con- sequences of the systems of beliefs arising from the social situations just mentioned are likely to appear in the form of unified group actions-that is, in the form of social movements aiming either to create social change or to preserve the status quo. We shall return later to an elaboration of the kinds of hypotheses that can be put forward con- cerning the creation of change versus the preser- vation of status quo. But before this is done, the realistic group conflict theory must be considered against this general background.

The implications of this conceptualization for intergroup relations in stratified societies and in- stitutions are both evident and direct. Whenever social stratification is based upon an unequal di- vision of scarce resources-such as power, pres- tige, or wealth-and hence there is a real conflict of interests between social groups, the social situ- ation should be characterized by pervasive ethno- centrism and out-group antagonism between the over- and underprivileged groups (Oberschall, 1973, p. 33). However, decades of research into ethnic-group relations suggest that ethnocentrism among stratified groups is, or at least it has been, very much a one-way street. Milner (1975,1981) and Giles and Powesland (1975) summarize a great deal of evidence that minority or subordinate group members-such as the American Blacks, the

280 • Political Psychology

French Canadians, the New Zealand Maoris, or the South African Bantus-have frequently tended to derogate the in-group and display positive atti- tudes toward the dominant out-group. In other words, deprived groups are not always ethnocen- tric in the simple meaning of the term; they may, in fact, be positively oriented toward the depriv- ing out-group. Data of this kind are not consistent with a simple application of RCT. (Recent detailed reviews of other field and laboratory data relevant to assessing the validity of the theory are provided by Brewer, 1979, Stephenson, 1981, and Turner, 1981.)

Some writers (including Gregor & McPherson, 1966; Milner, 1975, 1981; Morland, 1969) have argued that the status relations between dominant and subordinate groups determine the latter's iden- tity problems. (By social status we mean a rank- ing or hierarchy of perceived prestige.) Subordi- nate groups often seem to internalize a wider social evaluation of themselves as "inferior" or "second class," and this consensual inferiority is reproduced as relative self-derogation on a number of indices that have been used in the various studies. Con- sensual status itself-where subjective and ac- corded prestige are identical-is problematic for RCT, which conceptualizes prestige as a scarce resource, like wealth or power. Status differences between groups, like other inequalities, should tend to accentuate the intergroup conflict of interests. Therefore, according to RCT, the impact of low status upon a subordinate group should be to in- tensify its antagonism toward the high-status group (Thibaut, 1950). Yet, under some conditions at least, low social status seems to be cOITelated with an enhancement, rather than a lessening, of posi- tive out-group attitudes.

It could be argued that only conflicts of interest perceived as such create hostility. This requires that groups must compare their respective situations. And, according to some views, it is only relatively similar groups that engage in mutual comparisons; therefore, many forms of status differences will reduce perceived similarity (see Festinger, 1954; Kidder & Stewart, 1975). It follows that status systems may reduce social conflict by restricting the range of meaningful comparisons available to any given group. This hypothesis may be a useful tool to account for some of the determinants of social stability; but if it is taken to its logical con-

o elusion, it can account for no more than that. It

fails to account for social change (in the sen . changes in the mutual relations, behavior an~e of titudes of large-scale human groups that h~ve bat- distinctly different in status in the past) part·

een , ICU-

larly when the processes of change become v rapid. St~tus ~ifferences betv:een groups ofteneZ no~ remaIn uOllater~lly assOCiated with low levels of IOtergroup confliCt. For example, the generali_ zation made. above-th~t certain forms of politi- c~l, economIC, and s~cI~l subordination of a so- CIal grouP. tend ~o elImlOate or even reverse its ethnocentnsm-Is already dated. Research con- ducted over the last two decades reveals a chana _ ing pattern in intergroup relations. Americ: Blacks (Brigham, 1971; Friedman, 1969; Harri~' & Braun, 1971; Hraba & Grant, 1970), French Canadians (Berry, Kalin & Taylor, 1977), New Zealand Maoris (Vaughan, 1978) and the Welsh (Bourhis, Giles & Tajfel, 1973; Giles & Powesland 1975), for instance, now seem to be rejecting (0; have already rejected) their previously negative in- group evaluations and developing a positive eth- nocentric group identity. (Milner, 1981, and Tajfel, 1982b, argue that these new data are likely to be a genuine reflection of social change.) This construc- tion of positive in-group attitudes has often been accompanied by a new militancy over political and economic objectives (see Tomlinson, 1970).

But these developments do not rescue RCT in its original form. The very suddenness with which the scene has changed effectively rules out objec- tive deprivation and therefore new conflicting group interests as sufficient conditions for the "subordinate" group ethnocentrism. On the con- trary, there has often been less objective depriva- tion than there was in the past. An active and new search for a positive group identity seems to have been one of the critical factors responsible for the reawakening of these groups' claims to scarce re- sources (Dizard, 1970).

In summary, RCT states that opposing claims to scarce resources, such as power, prestige, or wealth, generate ethnocentrism and antagonism between groups. Therefore, low status should tend to intensify out-group hostility in groups that are politically, economically, or socially subordinate. The evidence suggests, however, that where so- cial-structural differences in the distribution of resources have been institutionalized, legitimized, and justified through a consensually accepted sta- tus system (or at least a status system that is suffi-

ciently firm and per~asive t.o prevent the creation f counitive alternatives to It), the result has been

~ess :nd not more et?nocent~ism in the different statUS groups. The pnce of thIS has often been the ubordinate group's self-esteem. On the other ~and, whenever a subordinate group begins, for whateVer reasons, to question or deny its presumed characteristics associated with its low status, this seems to facilitate the reawakening of a previously dormant conflict over objective resources. At the same time, it is likely that one of the counter- reactions from the dominant groups in such situa- tions will be to work for the preservation of the previous~y .existing "subjective" and "objective" differentIatIOns.

A tentative hypothesis about intergroup conflict in stratified societies can now be offered: An un- equal distribution of objective resources promotes antagonism between dominant and subordinate groupS, provided that the latter group rejects its previously accepted and consensually negative self-image, and with it the status quo, and starts working toward the development of a positive group identity. The dominant group may react to these developments either by doing everything possible to maintain and justify the status quo or by attempting to find and create new differentia- tions in its own favor, or both. A more detailed specification of some of the strategies and "solu- tions" that can be adopted in this situation can be found in Tajfel (1978a); we shall return later to a discussion of some of them. For the present, it will be sufficient to state that, whether valid or not, the hypothesis raises some important theoretical prob- lems that need to be considered. The first question is: What social-psychological processes are in- volved in the development of positive group iden- tity? The second question concerns the conditions under which the status differences between social groups are likely to enhance or to reduce inter- group conflict. In order to continue the discussion of these questions, we must now abandon specu- lation and consider some relevant data.

Social Categorization and Intergroup Discrimination

The initial stimulus for the theorizing presented here was provided by certain experimental inves- tigations of intergroup behavior. The laboratory

The Social Identity Theory of Intergroup Behavior • 281

a?alogue o~real-worldethnocentrism is in-group bIas-that IS, the tendency to favor the in-group over the out-group in evaluations and behavior. Not only are incompatible group interests not always sufficient to generate conflict (as concluded in the last section), but there is a good deal of experi- mental evidence that these conditions are not al- ways necessary for the development of competi- tion and discrimination between groups (Brewer, 1979; Turner, 1981), although this does not mean, of course, that in-group bias is not influenced by the goal relations between the groups.

All this evidence implies that in-group bias is a remarkably omnipresent feature of intergroup re- lations. The phenomenon in its extreme form has been investigated by Tajfel and his associates. There have now been in addition to the original studies (Tajfel, 1970; Tajfel et al., 1971) a large number of other experiments employing a similar procedure (methodological and conceptual issues concerning the experimental paradigm are dis- cussed by Aschenbrenner & Schaefer, 1980; Bornstein et al., 1983a; Bornstein et al., 1983b; Branthwaite, Doyle, & Lightbown, 1979; Brown, Tajfel, & Turner, 1980; Turner, 1980, 1983a, 1983b; and the results of the relevant studies are summarized most recently by Turner, 1983a, and in a wider theoretical and empirical context by Brewer, 1979; Brown & Turner, 1981; Turner, 1981, 1982), all showing that the mere perception of belonging to two distinct groups-that is, so- cial categorization per se-is sufficient to trigger intergroup discrimination favoring the in-group. In other words, the mere awareness of the pres- ence of an out-group is sufficient to provoke in- tergroup competitive or discriminatory responses on the part of the in-group.

In the basic paradigm the subjects (both chil- dren and adults have acted as subjects in the vari- ous studies) are randomly classified as members of two nonoverlapping groups-ostensibly on the basis of some trivial performance criterion. They then make "decisions," awarding amounts of money to pairs of other subjects (excluding self) in specially designed booklets. The recipients are anonymous, except for their individual code num- bers and their group membership (for example, member number 51 of the X group and member number 33 of the Y group). The subjects, who know their own group membership, award the amounts individually and anonymously. The re-

282 • Political Psychology

sponse format of the booklets does not force the subjects to act in terms of group membership.

In this situation, there is neither a conflict of interests nor previously existing hostility between the "groups." No social interaction takes place between the subjects, nor is there any rational link between economic self-interest and the strat- egy of in-group favoritism. Thus, these groups are purely cognitive and can be referred to as "minimal."

The basic and highly reliable finding is that the trivial, ad hoc intergroup categorization leads to in-group favoritism and discrimination against the out-group. Fairness is also an influential strategy. There is also a good deal of evidence that, within the pattern of responding in terms of in-group fa- voritism, maximum difference (MD) is more im- portant to the subjects than maximum in-group profit (MIP). Thus, they seem to be competing with the out-group, rather than following a strategy of simple economic gain for members of the in-group. Other data from several experiments also show that the subjects' decisions were significantly nearer to the maximum joint payoff (MlP) point when these decisions applied to the division of money between two anonymous members of the in-group than when they applied to two members of the out- group; that is, relatively less was given to the out- group, even when giving more would not have af- fected the amounts for the in-group. Billig and Tajfel (1973) have found the same results even when the assignment to groups was made explic- itly random. This eliminated the similarity on the performance criterion within the in-group as an alternative explanation of the results. An explic- itly random classification into groups proved in this study to be a more potent determinant of dis- crimination than perceived interpersonal similari- ties and dissimilarities not associated with catego- rization into groups. Billig (1973), Brewer and Silver (1978), Locksley, OItiz and Hepburn (1980), and Turner, Sachder and Hogg (1983) have all rep- licated this finding that even explicitly arbitrary social categorizations are sufficient for discrimi- nation, and Alien and Wilder (1975) have provided additional evidence for the importance of group classification compared to similarities between people without such classification.

The question that arises is whether in-group bias in these minimal situations is produced by some fotm of the experimenter effect or of the demand

characteristics of the experimental situation_' other words, whether explicit references to gr In

b h " oup

mem ers Ip commUnIcate to the subjects that th are expected to, or ought to, discriminate. The fire

y

. b d b h" stPOint to e ma e a out t IS IOterpretation of th results is that explicit references to group me ~ bership are logically necessary for operationalizi~ in these minimal situations the major independen~ vari~ble-thatis, social catego?zation per se. Thi~ requITes not merely that the subjects perceive them- selves as similar to or different from others as in- divid~als, b~t that they ar~ members of discrete and dlscontlOuo~s categone~-thatis, "groups." Second, a detailed analysIs of the subjects' postsession reports (Billig, 1972; Turner, 1975a) shows that they do not share any common con- ception of the "appropriate" or "obvious" way to behave, that only a tiny minority have some idea of the hypothesis, and that this minority does not always conform to it. Thirdly, the relevant experi- mental data do not support this interpretation. St. Claire and Turner (1982) exposed observer-sub_ jects to exactly the same experimental cues as nor- mal categorized subjects; the former were required to predict the responses of the latter in the stan- dard decision booklets. The categorized subjects did discriminate significantly, but the observers failed to predict it and in fact expected significantly more fairness than was actually displayed. . The more general theoretical problem has been referred to elsewhere by one of us as follows:

Simply and briefly stated, the argument (e.g., Gerard and Hoyt, 1974) amounts to the follow- ing: the subjects acted in terms of the intergroup categorization provided or imposed by the experi- menters, not necessarily because this has been successful in inducing any genuine awareness of membership in separate and distinct groups, but probably because they felt that this kind of be- havior was expected of them by the experiment- ers, and therefore they conformed to this expec- tation. The first question to ask is why should the subjects be expecting the experimenters to expect of them this kind of behavior? The Gerard and Hoyt answer to this is that the experimental situ- ation was rigged to cause this kind of expecta- tion in the subjects. This answer retains its plau- sibility only if we assume that what was no more than a hint from the experimenters about the no- tion of "groups" being relevant to the subjects' behavior had been sufficient to determine, pow- erfully and consistently, a particular form of in-

t raroup behavior. In turn, if we assume this- aenct the assumption is by no mean.s unre~son­ able-we must also assu~e .that thiS ~art~cular form of intergroup behavlOr IS one which is ca- pable of being induced by the experimenters much more easily than other forms (such as coopera- tion between the groups in extorting the maxi- mum total amount of money from the experiment- ers, or a fair division of the spoils between the aroups, or simply random responding). And this iast assumption must be backed up in its turn by another presupposition: namely, that for some reasons (whatever they may be) competitive be- havior between groups, at least in our culture, is extraordinarily easy to trigger off-at which point we are back where we started from. The problem then must be restated in terms of the need to specify why a certain kind of intergroup behav- ior can be elicited so much more easily than other kinds; and this specification is certainly not made if we rest content with the explanation that the behavior occurred because it was very easy for the experimenters to make it occur. (Tajfel, 1978a, pp. 35-36)

Two points stand out: first, minimal intergroup discrimination is not based on incompatible group interests; second, the baseline conditions for in- tergroup competition seem indeed so minimal as to cause the suspicion that we are dealing here with some factor or process inherent in the intergroup situation itself. Our theoretical orientation was developed initially in response to these clues from our earlier experiments. We shall not trace the his- tory of its development, however, but shall describe its present form.

Social Identity and Social Comparison

Many orthodox definitions of "social groups" are unduly restrictive when applied to the context of intergroup relations. For example, when members of two national or ethnic categories interact on the basis of their reciprocal beliefs about their respec- tive categories and of the general relations between them, this is clearly intergroup behavior in the everyday sense of the term. The "groups" to which the interactants belong need not depend upon the frequency of intermember interaction, systems of role relationships, or interdependent goals. From the social-psychological perspective, the essential criteria for group membership, as they apply to

The Social Identity Theory of Intergroup Behavior • 283

large-scale social categories, are that the individu- als concerned define themselves and are defined by others as members of a group.

We can conceptualize a group, in this sense, as a collection of individuals who perceive them- selves to be members of the same social category, share some emotional involvement in this com- mon definition of themselves, and achieve some degree of social consensus about the evaluation of their group and of their membership in it. Follow- ing from this, our definition of intergroup behav- ior is basically identical to that of Sherif (1967, p. 62): any behavior displayed by one or more actors toward one or more others that is based on the ac- tors' identification of themselves and the others as belonging to different social categories.

Social categorizations are conceived here as cognitive tools that segment, classify, and order the social environment, and thus enable the indi- vidual to undertake many forms of social action. But they do not merely systematize the social world; they also provide a system of orientation for self-reference: they create and define the individual's place in society. Social groups, un- derstood in this sense, provide their members with an identification of themselves in social terms. These identifications are to a very large extent re- lational and comparative: they define the individual as similar to or different from, as "better" or "worse" than, members of other groups. It is in a strictly limited sense, arising from these consider- ations, that we use the term social identity. It con- sists, for the purposes of the present discussion, of those aspects of an individual's self-image that derive from the social categories to which he per- ceives himself as belonging. With this limited con- cept of social identity in mind, our argument is based on the following general assumptions:

1. Individuals strive to maintain or enhance their self-esteem: they strive for a positive self-con- cept.

2. Social groups or categories and the member- ship of them are associated with positive or negative value connotations. Hence, social identity may be positive or negative according to the evaluations (which tend to be socially consensual, either within or across groups) of those groups that contribute to an individual's social identity.

3. The evaluation of one's own group is deter-

The Social Identity Theory of Intergroup Behavior • 285284 • Political Psychology

mined with reference to specific other groups through social comparisons in terms of value- laden attributes and characteristics. Positively discrepant comparisons between in-group and out-group produce high prestige; negatively discrepant comparisons between in-group and out-group result in low prestige.

From these assumptions, some related theoretical principles can be derived:

1. Individuals strive to achieve or to maintain posi- tive social identity.

2. Positive social identity is based to a large ex- tent on favorable comparisons that can be made between the in-group and some relevant out- groups: the in-group must be perceived as posi- tively differentiated or distinct from the relevant out-groups.

3. When social identity is unsatisfactory, individu- als will strive either to leave their existing group and join some more positively distinct group and/or to make their existing group more posi- tively distinct.

The basic hypothesis, then, is that pressures to evaluate one's own group positively through in- group/out-group comparisons lead social groups to attempt to differentiate themselves from each other (Tajfel, 1978a; Turner, 1975b). There are at least three classes of variables that should influ- ence intergroup differentiation in concrete social situations. First, individuals must have internal- ized their group membership as an aspect of their self-concept: they must be subjectively identified with the relevant in-group. It is not enough that the others define them as a group, although con- sensual definitions by others can become, in the long run, one of the most powerful causal factors determining a group's self-definition. Second, the social situation must be such as to allow for inter- group comparisons that enable the selection and evaluation of the relevant relational attributes. Not all between-group differences have evaluative sig- nificance (Tajfel, 1959), and those that do vary from group to group. Skin color, for instance, is apparently a more salient attribute in the United States than in Hong Kong (Morland, 1969); whereas language seems to be an especially sa- lient dimension of separate identity in French Canada, Wales, and Belgium (Giles & Johnson,

1981; Giles & Powesland, 1975). Third, in-grou do ~ot compare themselves with every cognitivel

s

available out-group: the out-group must be p y . d 1 . er-

celv~ ~s a re ev~nt c?mpans~n group. Similarity, prOXImIty, and sltuatlOnal salience are among th variables that determine out-group comparabilit e and pressures toward in-group distinctivene y,

. . ~ should 1I1crease as a functIOn of this comparabil_ ity. It is important to state at this point that, in man social situations, comparability reaches a muc~ wider range than a simply conceived "similarity" between the groups.

The aim of differentiation is to maintain or achieve superiority over an out-group on some dimensions. Any such act, therefore, is essentially competitive. Fully reciprocal competition between groups requires a situation of mutual comparison and differentiation on a shared value dimension. In these conditions, intergroup competition, which may be unrelated to the objective goal relations between the groups, can be predicted to occur. Turner (1975b) has distinguished between social and instrumental or "realistic" competition. The former is motivated by self-evaluation and takes place through social comparison, whereas the lat- ter is based on "realistic" self-interest and repre- sents embryonic conflict. Incompatible group goals are necessary for realistic competition, but mutual intergroup comparisons are necessary, and often sufficient, for social competition. The latter point is consistent with the data from the minimal group experiments that mere awareness of an out- group is sufficient to stimulate in-group favorit- ism, and the observations (Doise & Weinberger, 1973; Ferguson & Kelley, 1964; Rabbie & Wilkens, 1971) that the possibility of social com- parison generates "spontaneous" intergroup com- petition.

Social and realistic competition also differ in the predictions that can be made about the conse- quences for subsequent intergroup behavior of winning or losing. After realistic competition, the losing groups should be hostile to the out-group victors, both because they have been deprived of a reward and because their interaction has been ex- clusively conflictual. However, when winning and losing establish shared group evaluations concern- ing comparative superiority and inferiority, then, so long as the terms of the competition are per- ceived as legitimate and the competition itself as fair according to these terms, the losing group may

acquiesce in the .superiority of the winni~g o.ut- uroup. This ~cqUleSC~?Ce ~y ~ group consIden~g ~ elf as legitImately mfenor has been shown 111 ~t~dies by Caddick (1980, 1982), Commins and Lockwood, (1979) and Turner and Brown (1978). Several other studies .report ~ndi~gs that are in line with this interpretatIOn: 10sll1g m-groups do not always derogate, but. so~etimes upgrade, their evaluations of the w1l1nll1g out-g:oups (fo~ ex- ample, Bass & Dunteman, 1963; Wllson & MIller,

1961). Retrospectively, at least, the social-identity/so-

cial-comparison theory is consistent with many of the studies mentioned in the preceding section of this chapter. In particular, in the paradigm of the minimal group experiments, the intergroup dis- crimination can be conceived as being due not to conflict over monetary gains, but to differentia- tions based on comparisons made in terms of mon- etary rewards. Money functioned as a dimension of comparison (the only one available within the experimental design), and the data suggest that larger absolute gains that did not establish a dif- ference in favor of the in-group were sacrificed for smaller comparative gains, when the two kinds of gains were made to conflict.

There is further evidence (Turner, 1978a) that the social-competitive pattern of intergroup behav- ior holds even when it conflicts with obvious self- interest. In this study, the distribution of either monetary rewards or "points" was made, within the minimal intergroup paradigm, between self and an anonymous other, who was either in the in- group or in the out-group. As long as minimal con- ditions existed for in-group identification, the sub- jects were prepared to give relatively less to themselves when the award (either in points or in money) was to be divided between self and an anonymous member of the in-group, as compared with dividing with an anonymous member of the out-group. These results seem particularly impor- tant, since the category of "self," which is by no means minimal or ad hoc, was set here against a truly minimal in-group category, identical to those used in the earlier experiments. Despite this stark asymmetry, the minimal group affiliation affected the responses.

The theoretical predictions were taken outside of the minimal categorization paradigm in a fur- ther study by Turner (1978b). He used face-to-face groups working on a discussion task. In each ses-

sion, two three-person groups discussed an iden- tical issue, supposedly to gain an assessment of their verbal intelligence, and then briefly compared their respective performance. The subjects were 144 male undergraduates. The criterion for inter- group differentiation was the magnitude of in- group bias shown in the ratings of the groups' work. Half the triads, composed of Arts students, believed that verbal intelligence was important for them (High Importance); half, composed of Sci- ence students, did not (Low Importance). Half the sessions involved two Arts or two Science groups (Similar Out-group), and half involved one Arts and one Science group (Dissimilar Out-group). Finally, in the Stable Difference condition, sub- jects were instructed that Arts students were defi- nitely superior and Science students definitely in- ferior in verbal intelligence; in the Unstable Difference condition, there was no explicit state- ment that one category was better than the other. These variables were manipulated in a 2 x 2 x 2 factorial design.

The results showed that the Arts (High Impor- tance) groups were more biased than the Science (Low Importance) groups, that similar groups dif- ferentiated more than dissimilar groups in the Stable condition, but that they were no more bi- ased (and sometimes even less so) in the Unstable condition; and that, on some of the measures, there was a significant main effect for out-group simi- larity: in-group bias increased against a similar out- group. Although these data are relatively complex, they do support some of our theoretical expecta- tions and provide an illustration that variations in in-group bias can be systematically predicted from the social-identity/social-comparison theory.

We have argued that social and realistic compe- tition are conceptually distinct, although most of- ten they are empirically associated in "real life." In an experiment by Turner, Brown, and Tajfel (1979) an attempt was made to isolate the effects on intergroup behavior of the postulated autono- mous processes attributed to a search for positive social identity. Children were used as subjects, and the manipulations involved decisions by the sub- jects about the distribution of payments for par- ticipation in the experiment, to be shared equally by the in-group, between the in-group and the out- groups that were made relevant or irrelevant to comparisons with the in-group's performance. Monetary self-interest (of a magnitude previously

286 • Political Psychology The Social Identity Theory of Intergroup Behavior • 287

ascertained to be of genuine significance to the subjects) would have produced no difference in the distribution decisions involving the two kinds of out-group; it would also have led to decisions tending toward maximum in-group profit (MIP) rather than toward maximum difference (MD).

MD was the most influential strategy in the choices. Furthermore, when the subjects could choose in-group favoritism (MD + MIP) and/or a fairness strategy, they were both more discrimina- tory and less fair toward the relevant than the ir- relevant comparison group. Other measures of in- group favoritism produced an interaction between reward level and type of out-group: more discrimi- nation against the relevant than the irrelevant group with high rewards, and less with low rewards. Whatever may be other explanations for this in- teraction, we can at least conclude that when re- ward levels are more meaningful, in-group favor- itism is enhanced against a more comparable out-group, independently of the group members' economic interests. Indeed, insofar as the subjects used the MD strategy, they sacrificed "objective" personal and group gain for the sake of positive in-group distinctiveness.

A study by Oakes and Turner (1982) also de- serves mention here since it seems to provide some direct evidence for the social competition inter- pretation of the minimal group experiments. They simply compared the self-esteem of subjects cat- egorized as in Tajfel et al. (1971) but who were not asked to complete the decision booklets with subjects who were categorized and also discrimi- nated in the normal manner. The latter subjects were found to have higher self-esteem than the former-in line with the idea that discrimination serves to achieve a positive social identity. Need- less to say, work is progressing to replicate and explore this finding.

On the whole, the above studies provide some confirmation for the basic social-identity/social- comparison hypothesis. Further studies testing the theory in both field and laboratory settings and discussions of its application to the analysis of specific social contexts (e.g., male-female rela- tions, linguistic conflict, Protestant-Catholic con- flict in Northern Ireland, prejudice and black iden- tity, etc.) are to be found or are reviewed in Tajfel (1978b, 1982a, 1982b) and Turner and Oiles (1981). We shall now attempt to outline in general terms the analysis of inter-group behavior in strati-

fied societies implied by the theory when it is ap- plied to some of the problems raised in the seCond section.

Status Hierarchies and Social Change

The reconceptualization of social status attempted earlier needs now to be made more explicit. Sta- tus is not considered here as a scarce reSOurce Or commodity, such as power or wealth; it is the out- come of intergroup comparison. It reflects a group's relative position on some evaluative di. mensions of comparison. Low subjective status does not promote inter-group competition directly; its effects on inter-group behavior are mediated by social identity processes. The lower is a group's subjective status position in relation to relevant comparison groups, the less is the contribution it can make to positive social identity. The variety of reactions to negative or threatened social iden- tity to be discussed below are an elaboration of the principles outlined earlier in this chapter.

1. INDIVIDUAL MOBILITY

Individuals may try to leave, or dissociate them- selves from, their erstwhile group. This is prob- ably more likely the more they approach the "so- Cial mobility" pole of the continuum of belief-systems described previously. This strategy usually implies attempts, on an individual basis, to achieve upward social mobility, to pass from a lower- to a higher-status group. In a four-group hierarchy, Ross (1979) found a direct linear rela- tionship between low status and the desire to pass upward into another group. Many earlier studies report the existence of strong forces for upward social movement in status hierarchies. Tendencies to dissociate oneself psychologically from fellow members of low-prestige categories are known to many of us from everyday experience: they have been noted more systematically by lahoda (1961) and Klineberg and Zavalloni (1969), among oth- ers, and indirectly by the whole literature on ra- cial identification and preference. The most im- portant feature of individual mobility is that the low status of one's own group is not thereby changed: it is an individualist approach designed, at least in the short run, to achieve a personal, not a group, solution. Thus, individual mobility im-

I · s a disidentification with the erstwhile in-

pie grOUP'

2. SOCIAL CREATIVITY

The group members may seek p~sitive disti?ctive- ss for the in-group by redefimng or altenng the

n~ rnents of the comparative situation. This need ~:t involve any change in th~ gr?up's actual S?- ciaI position or access to O?Jectlve resources In elation to the out-group. It IS a group rather than ~n individualistic strategy that may focus upon:

(a) Comparing the in-group to the out-group on some new dimension. Lemaine (1966) found, for example, that children's groups that could not compare themselves favorably with others in terms of constructing a hut-because they had been assigned poorer building materials than the out-group-tended to seek out other dimensions of comparison involving new con- structions in the hut's surroundings. The prob- lems that obviously arise here are those of le- gitimizing the value assigned to the new social products-first in the in-group and then in the other groups involved. To the extent that this legitimization may threaten the out-group's superior distinctiveness, an increase in inter- group tension can be predicted.

(b) Changing the values assigned to the attributes of the group, so that comparisons which were previously negative are now perceived as posi- tive. The classic example is "black is beauti- ful." The salient dimension-skin color-re- mains the same, but the prevailing value system concerning it is rejected and reversed. The same process may underlie Peabody's (1968) finding that even when various groups agree about their respective characteristics, the trait is evaluated more positively by the group that possesses it.

(c) Changing the out-group (or selecting the out- group) with which the in-group is compared- in particular, ceasing or avoiding to use the high-status out-group as a comparative frame of reference. Where comparisons are not made with the high-status out-group, the relevant inferiority should decrease in salience, and self-esteem should recover. Hyman's (1942)

classic paper on the psychology of status sug- gested that discontent among low-status-group members is lessened to the degree that intraclass rather than intergroup comparisons are made. More recently, Rosenberg and Simmons (1972) found that self-esteem was higher among blacks who made self-compari- sons with other blacks rather than whites. Other work also suggests (see Katz, 1964; Lefcourt & Ladwig, 1965) that, in certain circumstances, black performance was adversely affected by the low self-esteem induced by the presence of the members of the dominant out-group. It follows that self-esteem can be enhanced by comparing with other lower-status groups rather than with those of higher status. This is consistent with the fact that competition be- tween subordinate groups is sometimes more intense than between subordinate and domi- nant groups-hence, for example, lower-class or "poor white" racism.

3. SOCIAL COMPETITION

The group members may seek positive distinctive- ness through direct competition with the out- group. They may try to reverse the relative posi- tions of the in-group and the out-group on salient dimensions. To the degree that this may involve comparisons related to the social structure, it im- plies changes in the groups' objective socialloca- tions. We can hypothesize, therefore, following RCT, that this strategy will generate conflict and antagonism between subordinate and dominant groups insofar as it focuses on the distribution of scarce resources. Data relevant to this strategy have been referred to earlier in this chapter.

Let us assume as an ideal case some stratifica- tion of social groups in which the social hierarchy is reasonably correlated with an unequal division of objective resources and a corresponding status system (based on the outcomes of comparisons in terms of those resources). Under what conditions will this not lead to intergroup conflict-or, more precisely, to the development of competitive eth- nocentrism on the part of the subordinate group?

First, to the extent that the objective and the subjective prohibitions to "passing" are weak (see our earlier discussion of the "social mobility" sys- tem of beliefs), low status may tend, in conditions

288 • Political Psychology

of unsatisfactory social identity, to promote the widespread adoption of individual mobility strat- egies, or at least initial attempts to make use of these strategies. Insofar as individual mobility implies disidentification, it will tend to loosen the cohesiveness of the subordinate group. This weak- ening of subjective attachment to the in-group among its members will tend: (a) to blur the per- ception of distinct group interests corresponding to the distinct group identity: and (b) to create obstacles to mobilizing group members for col- lective action over their common illlerests. Thus. the Iow morale that follows from negative social identity can set in motion disintegrative processes that, in the long run. may hinder a change in the group status.

Second, assuming that the barricrs (objective, moral, and ideological prohihitions) to leaving one's group are strong, unsatisfactory social iden- tity may stimulate social creativity that tends to reduce the salience of the subordinate/dominant group conflict of interest. Strategy 2(c) mentioned above is likely to be crucial here since. in general, access to resources such as housing, jobs, income, or education is sufficiently central to the fate of any group that the relevant comparisons are not easily changed or devalued. Few underprivileged groups would accept puverty as a \'il1ue. but it may appear more tolerable to the degree that compari- sons are made with even poorer groups rather than with those that are better ofT (see Runciman. 1966).

As noted above, some writcrs (Festinger, 1954; Kidder & Stewart, InS) imply that strategy 2(c) is a dominant response to status differences be- tween groups. The assumption is that intergroup comparability decreases as a direct function of perceived dissimilarity. If this were the whole story, then, somewhat paradoxically. the creation of a consensual status system would protect social iden- tity from invidious comparisons. The causal se- quence would be as follows: similar groups com- pare with each other; the outcome determines their relative prestige; the perceived status difference reduces their similarity and hence comparability; intergroup comparisons cease to be made; subjec- tive superiority and inferiority decrease in salience; correspondingly, the groups' respective self-es- teems return to their original point. There may be occasions when this social-psychological recipe for the maintenance of the status quo can be ob- served in something like its pure form. However,

we shall argue presently that there are many sta- tus differences that do not reduce comparability.

For the moment, we can note that both indi- vidual mobility and some forms of social creativ- ity can work to reduce intergroup connict over scarce resources-though with different implica- tions. The former is destructive of subordinate- group solidarity and provides no antidote to nega- tive social identity at a group level. The latter may restore or create a positive self-image but, it can be surmised, at the price either of a collective re- pression of objective deprivation or, perhaps. of spurious rivalry with some other deprived group. It is interesting in this context that the French Ca- nadians, having recently gained a more asserti\'e identity, are now apparently more disparaging of other minority groups than are the English Cana- dians (Berry et aI., 1977).

By reversing the conditions under which social stratification does not produce intergroup connict, we can hypothesize that negative social identity promotes subordinate-group competitiveness to- ward the dominant group to the degree that: (a) subjective identification with the subordinate group is maintained; and (b) the dominant group continues or begins to be perceived as a relevanl comparison group. As a great deal of work has been done in social psychology on the determi- nants of cohesiveness and loyalty within groups-· Hogg (1983), Turner et al. (1983). and Turner. Sachdev & Hogg (l983) have recently looked in particular at the problem of how groups that are associated with costs and deprivations (such as ~ubordinate ones) are able to maintain their cohe- siveness-we shall concentrate on the second con- dition.

Our hypothesis is that a status difference be- tween groups does not reduce the meaningfulness of comparison between them providing that there is a perception that it can be changed. For example, consider two football (or any other) teams that at the end of their season may have come first and second in their league respectively. There is no argument about which has the higher status, but alternative comparative outcomes were and, in the future, still will be possible. When the new season begins, the teams will be as comparable and com- petitive as they had been before. This example il- lustrates Tajfel's (1978a) distinction between se- cure and insecure intergroup comparisons. The crucial factor in this distinction is whether cogni-

th'e alternatives to the actual outcome are avail- able-whether other outcomes are conceivable. Status di fferences between social groups in social systems showing variou degrees of stratification can be distinguished in the same way. Whece sta- tu" relations are perceived as immutable. a part of the fixed order of things, then social identity is secure. It becomes insecure when the existing state of affairs begins to be questioned. An important corollary to this argument is that the dominanl or high-status groups, too, can experience in.<;ccure social identity. Any threat to the distincti\-cJy su- perior position of a group implies a potential loss of posilive comparisons and possible negau\'C com- parisons, which must be guarded against. Such a threat may deri\'e from the activity of the Iow-sta- tus group or from a conflict between the high-sta- tus group's OWI1 value system (for example. the sociopolitical morality) and the actual foundations of its superiority. Like low-status groups, the high- status groups will react to insecure social identity by scarching for enhanced group distinctiveness.

In brief. then. it is true that clear-cut status dif- ferences may lead to a quiescent social system in which neither the "inferior" nor the "superior" groups will show much ethnocentrism. But this "ideal type" situation must be considered in rela- tion to the perceived stability and Iegitimal.)' of the system. Perceived illegitimacy and/or instabil- ity provide new dimensions of comparabilitv that arc directly relevant to the attitudes and behavior of the social groups involved. whatever theirposi- tion in the system. This is the social-psychologi- cal counterpart to what is widely known today a~ "the revolution of rising expectations." Providing that individual mobility is unavailable or undesir- able, consensual inferiority will be rejected most rapidly when the situation is perceived as both unstable and illegitimate. This is (or was) prob- ably the set of conditions underlying the develop- ment of ethnocentrism among black Americans, French Canadians, and New Zealand Maoris, for instance. Vaughan (1978) reports that the perceived feasibility of social change (probably including, in this instance, the perceived illegitimacy of the present situation) is an important predictor of the developing Maori ethnocentl;sm; Friedman (1969) argues that what we may term the "cognitive al- ternative" of black nationalism in the developing countries was influential in enhancing black American social identity.

The Social Identity Theory of Intergroup Behavior • 289

On the other hand, when the dominant group or sections of it perceive their superiority as legiti- mate, they will probably react in an intensely dis- criminatory fashion to any attempt by the subor- dinate group to change the intergroup situation. Such perhaps was the postbellum situation in the southern United States: the whites, threatened by those \\'ho had been their slaves, rapidly abandoned their paternalistic stereotypes of the blacks as "childlike" in favor of openly hostile and deroga- lory ones (Van der Berghe, 1967). The reactions of illegitimately superior groups arc more com- plex (Turner & Brown, 1978). It seems that con- n icts of values are reduced by greater discri mi na- lion when superiority is assured. but by less discrimination when il is unstable. This calls to mind ~ollle Prisoner Dilemma studies in which white discrimination against black opponents in- creased the more cooperative was the opponent, but decreased the more competitive he was (Baxter, 1473: Cederblom & Diers, 1970). Baxter sug- gcsted in the title of his article ("Prejudiced Lib- erals')") that a conflict of values may underlie his data. Research on the different effects of secure and insecure status differences is reported in Tajfel (197Xb, 1982a. 1982b; see also Caddick, 1980 and Skevington, 1980).

Many of the points and hypotheses we have advanccd in this chaptcr are not, in themselves, nc\\' (sce, for inswnce, Sherif', 1967: Runcilllan, 1%6: Milner. 1975; Billig, 1976). What is new. wc think. is the integration of the three processes of social categorization, self-evaluation through social identity. and intergroup social comparison, into a coherent and testable framework for con- tributing to the explanation of various forms of intergroup behavior, social conflict, and social change. This framework contains possibilities of further development, and to this extent, we hope that it may stimulate theoretically directed research in areas that have not been considered here.

But some cautionary points should be made. The equation of social competition and intergroup con- flict made above rests on the assumptions concern- ing an "ideal type" of social stratification in which the salient dimensions of intergroup differentia- tion are those involving scarce resources. In this respect, we have simply borrowed the central te- net of RCT. There is no reason, in fact, to assume that intergroup differentiation is inherently conflictual. Some experimental work already

290 • Political Psychology The Social Identity Theory of Intergroup Behavior • 291

points clearly to the conclusion that evaluative derogation of an out-group is conceptually and empirically distinct from out-group hostility (Turner et al., 1979). On the other hand, social- identity processes may provide a source of inter- group conflict (in addition to the cases outlined above) to the degree that the groups develop con- flicting interests with respect to the maintenance of the comparative situation as a whole. It seems plausible to hypothesize that, when a group's ac- tion for positive distinctiveness is frustrated, im- peded, or in any way actively prevented by an out- group, this will promote overt conflict and hostility between the groups. This prediction, like many others, still remains to be tested.

'Objective' and 'Subjective' Conflicts

None of the arguments outlined in this chapter must be understood as implying that the social- psychological or "subjective" type of conflict is being considered here as having priority or a more important causal function in social reality than the "objective" determinants of social conflict of which the basic analysis must be sought in the social, economic, political, and historical structures of a society. The major aim of the present discus- sion has been to determine what are the points of insertion of social-psychological variables into the causal spiral; and its argument has been that, just as the effects of these variables are powerfully determined by the previous social, economic, and political processes, so they may also acquire, in turn, an autonomous function that enables them to deflect in one direction or another the subse- quent functioning of these processes.

It is nearly impossible in most natural social situations to distinguish between discriminatory intergroup behavior based on real or perceived conflict of objective interests between the groups and discrimination based on attempts to establish a positively valued distinctiveness for one's own group. However, as we have argued, the two can be distinguished theoretically, since the goals of actions aimed at the achievement of positively valued in-group distinctiveness often retain no value outside of the context of intergroup com- parisons. An example would be a group that does not necessarily wish to increase the level of its own salaries but acts to prevent other groups from get-

ting nearer to this level so that differentials are eroded. But the difficulty with this example~Ot with many other similar examples-is that, in th~

h . f l' IScase, t e preservatIOn 0 sa ary dIfferentials' probably associated with all kinds of objecti~S advantages that cannot be defined in terms o~ m?ney a.lone. In turn, some ~f these advantages wIll agam make sense only m the comparative framework of intergroup competition. Despite this confusing network of mutual feedbacks and inter- actions, the distinctions made here are important because they help us to understand some aspects of intergroup behavior that have often been ne- glected in the past.

A further distinction must be made between explicit and implicit conflicts-a distinction that has to do with conflicts that are "objective" in a different sense. A conflict may be "objective" de- spite the fact that the goals the groups are aiming for have no value outside of the context of inter- group comparison in that it may be institutional. ized and legitimized by rules and norms (of what- ever origin) accepted by the groups themselves. This was the case in Sherif's studies in their phase of competition between the groups; and it also is the case in any football match and in countless other social activities. The behavior toward out- groups in this kind of explicit conflict can be clas- sified, in turn, into two categories, one of which can be referred to as instrumental and the other as noninstrumental. The instrumental category con- sists of all those actions that can be directly re- lated to causing the group to win the competition. The noninstrumental category, which could be re- ferred to as "gratuitous" discrimination against the out-group, includes the creation of negative ste- reotypes and all other aspects of the "irrelevant" in-group/out-group differentiations so well de- scribed, for example, in Sherif's studies. The first category of actions is both commonsensically and theoretically accounted for by assuming nothing more than the group's desire to win the competi- tion-although this poses all the theoretical "com- parison" problems discussed in this chapter; the second category of actions can be directly and parsimoniously accounted for in terms of the so- cial-comparison/social-identity/positive-in-group- distinctiveness sequence described here.

The implicit conflicts are those that can be shown to exist despite the absence of explicit in- stitutionalization or even an informal normative

acceptance of their eXi~ten~e by t~e groups in- volved. The proof of theIr eXIstence IS to be found in the large number ~f st~dies .(and .also ev~r~day occurrences in real lIfe) m whIch dIfferentIatlO~s of all kinds are made between groups by theIr members although, on the face of it, there are no "reasons" for these differentiations to occur. Ex- amples of this have been provided in several stud- ies mentioned in this chapter in which the intro- duction by the subjects of various intergroup differentiations directly decreased the objective rewards that could otherwise have been gained by the in-group, or even directly by the individual. Findings of this kind, which can be generalized widely to many natural social situations, provide a clear example of the need to introduce into the complex spiral of social causation the social-psy- chological variables of the "relational" and "com- parative" kind discussed in this chapter.

REFERENCES

Adorno. T. w.. Frenkel-Brunswik, E., Levinson, D. 1., & Sanford, R. N. (1950). The aLtlhoritarian personality. New York: Harper. 1950.

Alien, V. L., & Wilder, D. A. (1975). Categorization, belief similarity, and intergroup discrimination. Joumal of Per- sonality & Social Psychology, 32(6),971-977.

Aschenbrenner, K. M., & Schaefer, R. E. (1980). Minimal intergroup situations: Comments on a mathematical model and on the research paradigm. European Joumal of Social Psychology, 10, 389-398.

Bass, B. M., & Dunteman, G. (1963). Biases in the evalua- tion of one's own group, its allies, and opponents. Joul'llal of Conflict Resolwion, 2,67-77.

Baxter, G. W. (1973). Prejudiced liberals? Race and informa- tion effects in a two-person game. Joumal ofConflict Reso- lution, 17, 131-161.

Berkowitz, L. (1962). Aggression: A social psychological analysis. New York: McGraw-Hill.

Berko~itz, L. (1969). The frustration-aggression hypothesis revisited. In L. Berkowitz (Ed.), Roots of aggression: A re- examination of the frustration-aggression hypothesis. New York: Atherton Press.

Berkowitz, L. (1974). Some determinants of impulsive ag- gression: Role of mediated associations with reinforcements for aggression. Psychological Review, 81, 165-176.

Berry, 1. W., Kalin, R., & Taylor, D. M. (1976). Multiculturalism and ethnic aUill/des in Canada. Kingston, Ontario: Queen's University.

Billig, M. (1972). Social categorization in illlergroup rela- tions. University of Bristol, Bristol.

Billig, M. (1973). Normative communication in a minimal intergroup situation. European Journal of Social Psychol- ogy, 3(3), 339-343.

Billig, M. (1976). Social psychology GIld intergroup relatiolls (Vo!. 9). London: Academic Press.

Billig, M., & Tajfel, H. (1973). Social categorization and simi-

larity in intergroup behaviour. Europeall Journal of Social Psychology, Vo!. 3(1),27-52.

Blake, R. R., & Mouton, 1. S. (1961). Competition, commu- nication and conformity. In 1. A. Berg & B. M. Berg (Eds.), Conformity and deviation. New York: Harper.

Blake, R. R., & Mouton, 1. S. (1962). The intergroup dynam- ics of win-lose conflict and problem-solving collaboration in union-management relations. In M. Sherif (Ed.), Imer- glVup relations and leadership. New York: Wiley.

Bornstein, G., Crum, L., Wittenbraker, 1., Harring, K., Insko, C. A., & Thibaut, 1. (1983a). On the measurement of social orientation in the minimal group paradigm. European Jour- nal of Social Psychology, 13,321-350.

Bornstein, G., Crum, L., Wittenbraker, 1., HalTing, K., Insko, C. A., & Thibaut, 1. (l983b). Reply to Turner's comments. European Journal of Social Psychology, 13, 360-381.

Bourhis, R. Y., Giles, H., & Tajfel, H. (1973). Language as a determinant of Welsh identity. European Journal of Social Psychology, 3,447-460.

Brantwaite, A., Doyle, S., & Lightbown, N. (1979). The bal- ance between fairness and discrimination. European Jour- nal of Social Psychology, 9, 149-163.

Brewer, M. B. (1979). In-group bias in the minimal intergroup situation: A cognitive-motivational analysis. Psychologi- cal Bulletin. 86(2), 307-324.

Brewer, M. B., & Silver, M. (1978). Ingroup bias as a func- tion of task characteristics. European Journal of Social Psychology, 8(3), 393-400.

Brigham, 1. C. (1971). Views of White and Black schoolchil- dren conceming racial differences. Paper presented at the Midwestern Psychological Association, Detroit, Michigan.

Brown, R.l., Tajfel, H., & Turner, 1. C. (1980). Minimal group situations and intergroup discrimination: Comments on the paper by Aschenbrenner and Schaefer. European Joumal of Social Psychology, 10(4),399-414.

Brown, R. J., & Turner, J. C. (1981). Interpersonal and inter- group behaviour. In J. C. Turner & H. Giles (Eds.), IlIter- group behaviour. Oxford: Basil Blackwell.

Caddick, B. (1980). Equity theory, social entity and inrergroup relations. Review of Personaliry alld Social Psychology, 1, 219-245.

Caddick, B. (1982). Perceived illegitimacy and intergroup relations. In H. Tajfel (Ed.), Social identity and illtergroup relatiol/s. Cambridge: Cambridge University Press.

Campbell, D. T. (1965). Ethnocentric and other altruistic motives. In D. Levine (Ed.), Nebraska Symposium Oil Mo- tivation (pp. 283-311). Lincoln: Nebraska: University of Nebraska Press.

Cederblom, D., & Diers, C.l. (1970). Effects of race and strat- egy in the prisoner's dilemma. Journal of Social Psychol- ogy, 81, 275-276.

Commins, B., & Lockwood, 1. (1979). The effects of status differences, favoured treatment, and equity on intergroup comparisons. Europeall Journal of Social Psychology, 9, 281-290.

Deutsch, M. (1949). A theory of cooperation and competi- tion. Human Relations, 2, 129-151.

Diab, L. (1970). A study of intragroup and inrergroup rela- tions among experimentally produced small groups. Ge- netic Psychology Monographs, 82, 49-82.

Dizard, 1. E. (1970). Black identity, social class, and Black power. Psychiatry, 33, 195-207.

The Social Identity Theory of Intergroup Behavior • 293292 • Political Psychology

Doise, W., & Weinberger, M. (1973). Representations masculines dans differentes situations de rencontres mixtes. Bulletin de Psychologie, 26,649-657.

Ferguson, C. K., & Kelley, H. H. (1964). Significant factors in overevaluation of own group's products. Journal of Ab- normal & Social Psychology, 69, 223-228.

Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117-140.

Fiedler, F. E. (1967). The effect of inter-group competition on group member adjustment. Personnel Psychology, 20( I), 33-44.

Friedman, . (1969). Africa and the Afro-Americans: The changing Negro identity. PsychiallY, 32(2), 127-136.

Giles, H., & Johnson, P. (1981). The role of language in eth- nic group relations. In J. C. Turner & H. Giles (Eds.), In- rergroup behavior. Oxford: Basil Blackwell.

Giles, H., & Powesland, P. E. (1976). Speech style and social evaluarions. London: Academic Press, European Mono- graphs in Social Psychology.

Gregor, A. J., & McPherson, D. A. (1966). Racial preference and ego identity among White and Bantu children in the Republic of South Africa. Generic Psychology Monographs, 73,217-254.

Harris, S., & Braun, J. R. (1971). Self-esteem and racial pref- erences in Black children. Proceedings of rhe 79th Annual Convention of the American Psychological Association, 6.

Harvey, O. J. (1956). An experimental investigation of nega- tive and positive relations between small groups through judgmental indices. Sociomerry, 19,201-209.

Hirschman, A. O. (1970). Exit, voice and loyalry: Responses to decline in firms, organizations and states. Cambridge, MA: Harvard University Press.

Hogg, M. A. (1983). The social psychology of group-Jorma- tion: A cognitive perspecrive. Unpublished doctoral disser- tation, University of Bristol.

Hraba, J., & Grant, G. (1970). Black is beautiful: A reexami- nation of racial preference and identification. JOLlrnal of Personaliry & Social Psychology, 16(3),398-402.

Jahoda, G. (1961). White man. Oxford, England: Oxford Uni- vel'S ity Press.

lohnson, D. W. (1967). Use of role reversal in intergroup com- petition. Journal of Personality & Social Psychology, 7(2), 135-141.

Julian, J. W. (1968). The study of competition. In W. E. Vinacke (Ed.), Readings in general psychology. New York: American Book Company.

Kalin, R., & Marlowe, D. (1968). The effects of intergroup competition, personal drinking habits and frustration in intra-group cooperation. Proceedings ofthe American Psy- chological Association, 3, 405-406.

Katz, 1. (1964). Review of evidence relating to effects of de- segregation on the intellectual performance of Negroes. American Psychologist, 19(6), 381-399.

Kidder. L. H., & Stewart, V. M. (1975). The psychology of intergroup relations. New York: McGraw-Hill.

Klineberg, 0., & Zavalloni, M. (1969). Nationalism and tribal- ism among African students. The Hague and Paris: Mouton.

Lefcourt, H. M., & Ladwig, G. (1965). The effect of refer- ence group upon Negroes' task persistence in a biracial competitive game. Journal of Personality and Social Psy- chology, I, 688-671.

Lemaine, G. (1966). Inegalite, comparison et incomparabilite:

Esquisse d'une theorie de l'originalite socialite. Bulle . de Psychologie, 252(20), 1-2,1-9. tin

Locksley, A., Ortiz, v., .& ~epburn, C. (1980). Social Cate 0- nzallon and dlscnmmatlon behaVIOur: ExtinguiShin gh minimal intergroup discrimination effect. Joul'/lal oJR

t e

sonaliry and Social Psychology, 39, 773-783. er· Miln.er, D. (1975). Children and race. HarmondsWOrth

Middlesex: Pengum. ' Milner, D. (1981). Racial prejudice. In l. C. Turner & H. Giles

(Eds.), IntergroLlp behavior. Oxford: Basil Blackwell. Morland, 1. K. (I ?69). Race awareness among American and

Hong Kong Chmese chIldren. American Journal ofSociol. ogy, 75(360-374).

Oakes, P. J., & Turner, J. C. (1982). Social categorization and intergroup behavio~r: Does minima~ ~ntergroup diScrimi_ natIOn make SOCial Identity more pOSitive? European Jour- nal of Social Psychology, 10, 295-30 l.

Oberschall, A. (1973). Social conflict and social movements New York: Prentice-Hall. .

Peabody, D. (1968). Group judgments in the Philippines: Evaluative and descriptive aspects. Journal of Personality and Social Psychology, 10,290-300.

Rabbie, J. M., & Wilkens, C. (1971). Intergroup competition and its effects on intra- and intergroup relations. European Journal of Social Psychology, 1, 215-234.

Rosenberg, M., & Simmons, R. G. (1972). Black and White self-esteem: The urban school child. Unpublished manu- script.

Ross, G. F. (1979). Multiple group membership, social mo- biliry and intergroup relations. Unpublished doctoral dis- sertation, University of Bristol.

Runciman, W. G. (1966). Relative deprivation and sodaljlls, tice. London: Routledge and Keegan Paul.

Sherif, M. (1967). Social interaction: Process and prodllcts. Sherif, M., Harvey, O. l., White, B. l., Hood, W. R., & Sherif,

C. W. (1961).lntergroup cooperation and competition: The Robber's Cave experimelll. Norman, OK: University Book Exchange.

Sherif, M., & Sherif, C. W. (1953). Groups in harmony and tension; an integration of studies of intergroup relations. New York: Harper.

Skevington, S. M. (1980). Intergroup relations and social change within a nursing context. British Journal of Social & Clinical Psychology, 19,201-213.

St Claire, L., & Turner, J. C. (1982). The role of demand char- acteristics in the social categorization paradigm. Europeall Journal of Social Psychology, 12(3),307-314.

Stephenson. G. M. (1981). Intergroup bargaining and nego- tiation. In J. C. Turner & H. Giles (Eds.), Inrergrollp be- ha viol'. Oxford: Basil Blackwell.

Tajfel, H. (1959). A note on Lambert's "Evaluational reac- tions to spoken languages." Canadian Journal of Psychol- ogy, 13, 86-92.

Tajfel, H. (1970). Experiments in intergroup discrimination. Scientific American. 223(5), 96-102.

Tajfel, H. (1972a. Experiments in a vacuum. In J. Israel & H. Tajfel (Eds.), The context of social psychology: A critical assessment. London: Academic Press, European Mono- graphs in Social Psychology.

Tajfel. H. (1972b. La categorisation sociale. In M. S (Ed.), Introdllction iz la psychologie sociale (Vol. I). Paris: Larousse.

a'fel H. (1974a). Intergroup behaviOl; social comparison T ~Ild social change. Paper presented at the Katz-Newcomb

Lectures, University of Mlch~gan, Ann Arbor. . Tajfel. H. (1974b). SOCIal Identity and Intergroup behavIOUr.

social Science Informal/on, 1~(2), 65~93. . T .~ I H. (1975). The exit of SOCIal mobIlIty and the vOice of a~:Cial change: Notes on the social psychology of intergroup relations. Social Sciencelnfornll/tion. 14(2),.101-118 ..

Tajfel, H. (1978a). The achleve.ment of group dlffer.enuatlOn, DifferentiatIOn berween soclOl groups: SlUdles II! rhe so- cial psychology of intergrollp relations. London: Academic

Press. T 'fel, H. (1978b). The psychological structure of intergroup a~elations. In H. Tajfel (Ed.), Differentiation berween social gIVlIPS: Studies in the so~ial psychology of intergroup re- lations. London: AcademiC Press.

Tajfel, H. (1981). Human groups and social categories. Cam- bridge, UK: Cambridge University Press.

Tajfel, H. (1982a). Social psychology of intergroup relations. Allnl/al Re\'iew of Psychology, 33, 1-39.

Tajfel, H. (Ed.). (1982b). Social identiry and intergroup rela- tions. Cambridge: Cambridge University Press.

Tajtel, H., & B illig, M. (1974). Familiarity and categoriza- tion in intergroup behavlOr. Journal of Experunental So- cial Psychology, Vol. J0(2), 159-170.

Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. (1971). Social categorization and intergroup behaviour. European JOl/rnal of Social Psychology, Vol. 1(2), 149-178.

Thibaut, J. (1950). An experimental study of the cohesive- ness of underpriviledged groups. HI/man Relations, 3,251- 278.

Tomlinson, T. M. (1970). Contributing factors in Black poli- tics. Psychiatly, 33(2), 137-281.

Turner, J. C. (l975a). Social categori~ation of social com- parison in inrergroup relations. Unpublished doctoral dis- sertation, University of Bristol.

Turner. J. C. (1975b). Social comparison and social identity: Some prospects for intergroup behaviour. EI/ropean Jour- Ilal of Social Psychology, 5(1), 5-34.

Turner, J. C. (1978a). Social categorization and social dis- crimination in the minimal group paradigm. In H. Tajfel (Ed.), Differelltiation between social grollps: Studies in the social psychology of intergroup relarions (pp. 101-140). London: Academic Press.

Turner. J. C. (I 978b). Social comparison, similarity and ingroup favoritism. In H. Tajfel (Ed.), Differentiation be- rween social groups: Studies in the social psychology of intergroup relations (pp. 235-250). London: Academic Press.

Turner, J. C. (1980). Fairness or discrimination in intergroup

behavior? A reply to Branthwaite, Doyle and Lightbown. European JOl/rHal of Social Psychology, 10(2), 131-147.

Turner. J. C. (1981). The experimental social psychology of intergroup behavior. In J. C. Turner & H. Giles (Eds.), In- tergroup behavior. Oxford: Basil Blackwell.

Turner. J. C. (1982). Towards a cognitive redefinition of the social group. In H. Tajfel (Ed.), Social identiry and inter- group relarions (pp. 15-40). New York: Cambridge Uni- versity Press.

Turner, J. C. (1983a). A second reply to Bornstein, Crum, Wittenbraker, Harring, Insko and Thibaut on the measure- ment of social orientations. European JOl/mal of Social Psychology, 13(4),383-387.

Turner, J. C. (1983b). Some comments on ... "the measure- ment of social orientations in the minimal group paradigm." European JOl/mal of Social Psychology, 13(4),351-367.

Turner, J. C. (1984). Social identification and psychological group formation. In H. Tajfel (Ed.), The social dimension: European developments in social psychology. New York: Cambridge University Press.

Turner, J. C., & Brown, R. (1978). Social status, cognitive alternatives and intergroup relations. In H. Tajfel (Ed.), Differentiation between social groups: Sw dies in the so- cial psychology of intergroup relations (pp. 201-234). New York: Academic Press.

Turner, J. c., & Brown, R. J. (1976). Social status, cognitive alternatives and intergroup relations. In H. Tajfel (Ed.), Differentiation between social groups: Srudies in the social psychology of intergroup relations. London: Academic Press.

Turner, J. c., Brown, R. J., & Tajfel, H. (1979). Social com- parison and group interest in ingroup favouritism. Euro- pean JOl/mal of Social Psychology, 9(2), 187-204.

Turner, J. C., & Giles, H. (Eds.) (1981). Intergroup behavior. Oxford: Basil Blackwell.

Turner, J. c., Sachdev, I., & Hogg, M. A. (1983). Social cat- egorization, interpersonal attraction and group formation. British JourHal of Social Psychology, 22(3),227-239.

Van Den Berghe, P. L. (1967). Race and racism. New York: Wiley.

Vaughan, G. M. (1978). Social change and intergroup prefer- ences in New Zealand. European Jourtlal of Social Psy- chology, 8,297-314.

Vinacke, W. E. (1964). Intra-group power differences, strat- egy, and decisions in inter-triad competition. Sociomerry, 27,27-40.

Wilson, w., & Miller, N. (1961). Shifts in evaluations of par- ticipants following intergroup competition. Joumal of Ab- normal & Social PS.'ichology, 63,428-431.