Most of the theoretical and empirical work on social discrimination refers to Gordon Allport (1954), who defines discrimination as behavior that “. . . comes about only when we deny to individuals or groups of people equality of treatment which they may wish”
(p. 51). This definition comprises two crucial aspects: First, social discrimination neces- sitates social categorization; and second, it involves not only an actor, that is, somebody deciding about and realizing an intergroup treatment, but also a recipient, who disagrees with this treatment.
A more recent definition by Dovidio and collaborators describes social discrimination as “inappropriate treatment to individuals due to their group membership” (Dovidio, Brigham, Johnson, & Gaertner, 1996, p. 279). By reference to the issue of inappropri- ateness, this statement illustrates a further decisive aspect: Processes of judgment and interpretation, not just clear-cut, “objective” characteristics of the intergroup treatment determine what we conceive of as social discrimination (Otten & Mummendey, 1999a).
Correspondingly, Graumann and Wintermantel (1989) state that “the concept of social discrimination is inextricably connected to notions of justice and equ(al)ity . . .” (p. 183).
Instances of discrimination are not only restricted to category-based transgressions of the equality-norm, but can also imply transgressions of the equity-norm. Neither do equal distributions of resources guarantee that there is nodiscrimination, nor is unequal inter- group allocation a reliable criterion to identify instances of bias, favoritism, and deroga- tion. A key criterion is some dissent between allocating group (member) and targeted group (member) about the appropriateness of their relative treatment. Accordingly, the subsequent analysis will focus upon the different facets that might determine such dissent.
Differentiation versus discrimination
Social differentiation does not necessarily imply social discrimination, but forms the basis for both consensual and conflictual forms of unequal intergroup treatment. Thus, it is a necessary, but not a sufficient condition for social discrimination. As such, it can take dif- ferent forms: First, it can refer to a quantitativedimension – for example, those who work longer get more salary. Second, category-based treatment can follow a qualitativediffer- entiation – for instance, more money for those with better qualifications. Third, social differentiation can determine the access to opportunities and rights by reference to descrip- tive, inherited characteristics – for example, giving the right to vote only to citizens of 18 years and above. In all of these cases the parties involved can be demarcated into those who get more and those who get less, into those who have access to a certain resource and those who do not. None the less, for the examples given, we can also assume a shared consensus about the adequacy of the criteria applied or the dimension underlying differ- entiation. However, if there is dissent about the categorization that has been applied, or about the consequences derived from the categorical distinction, then social differentia- tion will turn into social discrimination.
Historical and cultural differences. Depending on the social context, the identical differ- entiation might be perceived as adequate and functional or as illegitimate and aversive.
Such contextual differences can be due to historical change. For example, the changed ideas on gender equality demonstrate the inter-time perspective on the (in)appropriate- ness of social differentiation: Until the 20th century women were typically conceived of as predisposed to raise children, keep the house, feed their family, while men were obliged to provide for their families materially. Gender-specific roles were assumed to fit gender- specific abilities (see Eagly, 1987). Even in democratic systems, it took a long time until the claim “Women and men are equal!” was heard and translated into social reality. For example, not until the 1960s were women accepted as students at prestigious American universities such as Princeton. Today, however, the same universities would be sued if they rejected a student based on gender. In addition to such variations over time, there are also examples of diverging inter-cultural perspectiveson certain categorical distinctions (see Triandis & Trafimow, this volume, chapter 18, for a discussion of cultural differences).
Discrimination and social categorization
Social discrimination can involve individual group members or groups as a whole. What makes it a socialphenomenon is not the number of persons involved but the fact that it presupposes social identities and is based upon categorical distinctions among groups of individuals. As outlined in the previous sections, the distinction of individuals in terms of social categories is not problematic per se, but its application as a basis of resource allo- cations might cause conflict and dissent. Accordingly, a crucial question for the social psy- chological analysis of social discrimination is: Why does a certain categorization of persons into members of ingroup and outgroup become the adequate basis for their differential treatment? Mere category salience does not necessarily imply a differential treatment according to the lines of this categorization. For example, article 3, 3 of the German con- stitution explicitly says: Nobody should be (dis)advantaged because of gender, descent, race, language, origin, faith, religious or political beliefs. Although these are highly acces- sible, frequently applied social categories, their members have to be treated equally before the law, and are guaranteed equal opportunities.
Mere categorization effects. Although salient social categories must not imply unequal treatment based on this distinction, a huge body of research on so-called minimal groups (Rabbie & Horwitz, 1969; Tajfel, Billig, Bundy, & Flament, 1971) documents that simply by introducing a categorization into “we” and “they,” even when based on fully arbitrary criteria and without directly serving realistic self-interest, ingroup bias and intergroup dis- crimination occur (e.g., Brewer, 1979; Brown, 1999). “Social identity theory” (Tajfel &
Turner, 1986) offered an account for this “mere categorization effect” by emphasizing the motivational functions of social categorization. People’s desire to see themselves positively can partly be derived from their membership of social groups (Tajfel, 1981); by perceiv- ing and treating own groups in a way that distinguishes them positively from other groups, individuals can enhance or ensure their positive social identity. Thus, even positive dif- ferentiation of a minimal ingroup can serve an overall positive self-concept.
Social categorization and identification. According to Tajfel and Turner (1986), identifi- cation with the ingroup is necessary for the assumed sequence from categorization via social comparison to positive ingroup distinctiveness. Research showed, however, that when measuring ingroup identification and intergroup bias, especially under minimal conditions, correlations are often only weak (see Hinkle & Brown, 1990). More recently, ingroup identification has been manipulated as an independent variable (Perreault &
Bourhis, 1999). Here, the relation between identification and favoritism turned out to be straightforward: Under conditions increasing ingroup-identification (chosen group membership) discriminatory behavior was significantly stronger than under conditions with lower identification (assigned minimal group membership). Other studies manipu- lating or measuring ingroup identification independently of the intergroup treatment similarly reveal that ingroup-identification can be understood as a key variable of discriminatory behavior (e.g., Branscombe & Wann, 1994; Ellemers, Van Rijswijk, Roefs,
& Simons, 1997).
Police officers and police dogs: Defining adequate levels of categorization. It is beyond the scope of this chapter to deal with empirical and theoretical work criticizing and extend- ing the theoretical explanation of the “mere categorization effect” as provided by social identity theory (e.g., Diehl, 1989; Mummendey, 1995). In the present chapter we can state that understanding social discrimination as a result of positive distinctiveness striv- ing cannot suffice to explain why certain category distinctions are taken as rationales for (unequal) distributions of resources while others are not. Here, self-categorization theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987) allows more precise predictions: An individual will apply a certain social categorization, if it subjectively gives meaning to the given situation. Such meaning and, thus, category salience is determined by three aspects:
accessibility, structural fit, and normative fit (Oakes, 1987). Accessibility refers to the per- ceiver’s readiness to apply a certain categorization, which is due to prior experiences or to situational goals. Structural fit is defined by the meta-contrast ratio, the proportion of perceived intra-category differences and inter-category differences; the more the latter outscore the former, the higher is the structural fit (Turner et al., 1987). Finally, norma- tive fit refers to the match between category and the content properties of its stimuli (cf.
Oakes, 1996), that is, it indicates whether in a given situation a categorization in fact allows appropriate behavioral predictions (e.g., in a decision about building of a new atomic reactor, the members of the Christian Democrats might predominantly stress eco- nomic interests while members of the Green Party will mainly focus on environmental risks).
The notion of fit, however, presupposes a general comparability of the two categories.
In order to guide intergroup treatment a social categorization must be linked to a super- ordinate, inclusive category defining a frame for the comparison. Consider a member of the police wondering whether his salary is fair. He might compare salaries of members of the secret service with those from the mounted police, but not consider relevant what the State spends for its human police members compared to its police dogs. A dissent about the intergroup treatment necessitates that there is an inclusive category (e.g., human police members) within which ingroup (e.g., secret service) and outgroup (e.g., mounted police) can compete for their entitlements in a certain allocation decision.
Categorization and justice. We have argued that discrimination is inextricably linked to both justice and social categorization (see also Tyler, this volume, chapter 17). Both aspects are taken into account in an approach by Wenzel (1997; 2000) who combined self-categorization theory (Turner et al., 1987) with theories of distributive justice.
According to self-categorization theory, ingroup favoritism implies that the own group is seen as more prototypically representing dimensions that are defined as crucial for the superordinate category including both ingroup and outgroup (Turner et al., 1987;
Mummendey & Wenzel, 2000). For example, a male employer could argue: Jobs in busi- ness necessitate advanced skills in math. Males are better at math than females; thus, male applicants should be preferred. Wenzel (1997; in press) argues and provides evidence that perceptions of social justice follow the same logic: Perceived entitlements are bound to the positive or negative evaluation of the own subcategory compared to the correspond- ing other subcategory in terms of relevant dimensions of the primary category, which includes all potential recipients in the given allocation situation.
Crucial is the social categorization of the target entity: As one possibility, the target may be categorized as a member of the category that includes the potential recipients in a certain allocation situation. In this case, the target is perceived to be equal to the other members of the primary category and, consequently, entitled to the same treatment. If, however, the primary category is further divided into subcategories (e.g., gender), then the subgroups’ prototypicality in terms of comparison dimensions relevant for the primary category (e.g., math skills) will determine perceived entitlement. Thus, we can conclude that salience and abstraction level of a certain categorization crucially determine whether social inequality will be perceived, accepted, neglected, questioned, or protested against (see also Major, 1994).
Discrimination as perspective-specific
As mentioned above, Allport’s (1954) definition of the term “social discrimination”
involves the issue of perspective. None the less, until recently, the overwhelming body of social-psychological theorizing and research on social discrimination has focused on the agents of social discrimination. Such one-sided focus on an interactionalphenomenon is not a deficiency unique to the domain of intergroup research, but has also been criticized in traditional aggression research (e.g., Mummendey, Linneweber, & Löschper, 1984;
Mummendey & Otten, 1993; Tedeschi, 1984). While recently there has been a growing interest in the consequences of social inequality and social discrimination on the targets’
side (see Crocker & Quinn and Ellemers & Barreto, this volume, chapters 12 and 16, respectively), the relation between target and actor still lacks a thorough theoretical empirical investigation.
Specific to the perspectives of either allocating/privileged group (member) or receiv- ing/disadvantaged group (member) the single groups’ entitlement to the resources in ques- tion will be estimated. As outlined in the previous section, social categorization provides a basis for judging this entitlement: Those who are categorized as equal or prototypical within the primary category of recipients deserve equal treatment, while those who are categorized as unequal may legitimately be treated unequally (Wenzel, 1997). Accord-
ingly, a perspective-specific dissent can involve three aspects: (1) Is the division of the primary category into subcategories appropriate (e.g., should job candidates be distin- guished in terms of gender)? (2) Is the value differentiation linked to the categorization appropriate (e.g., is it true that women are worse in math than men)? (3) Are the deci- sions about intergroup treatment derived from categorization and value differentiation appropriate (e.g., does the job require high proficiency in math)? In other words, social discrimination as an interactional phenomenon is characterized by a lack of consensus about the fit (Oakes, 1987) of the given social categorization in the respective social context.
As outlined above, an unequal or negative treatment of a group per se does not yet define an instance of social discrimination, but it might be accepted as finally functional, or as legitimized by different needs or different inputs on behalf of the differentiated groups. Groups might be willing to take serious burdens without complaint. Relative deprivation theories (see Taylor & Moghaddam, 1994) state that it is not hardship and negative living conditions per se, which make people feel unjustly treated: The key point is the experience that there are others who are better off, although they are perceived as equal in terms of entitlement or deservingness (see Mikula, 1994 for a respective analy- sis in interpersonal relations).
However, acknowledging perspectivity in social discrimination and defining a judg- mental dissent as its manifesting characteristic might provoke a serious misunderstand- ing. Our position does not imply that we do not have to care about the exploitation of groups so long as they do not complain about their fate. First, the dissent may not only be located between target and allocating party, but also between an outside observer and the allocating party. Second, such conclusion would imply a confound between the analy- sis of social discrimination from a political point of view with our attempt to clarify the psychological processesthat are at stake for the individuals involved in specific instances of implementing or receiving socially discriminating intergroup treatment. Only by analyz- ing how bothallocators and recipients differ in their understanding of the appropriate- ness of certain categorizations, value differentiations and, finally, intergroup allocations, can social discrimination as an interactional phenomenon be fully understood and con- vincingly dealt with.
Valence, behavioral mode, and discriminating behavior
The reflection on the distinction between differentiation and discrimination already implied that the quality of the “treatment” which might be perceived as discrimination needs a careful analysis. It is not sufficient to take into account whether a certain evalu- ation or allocation violates norms of equality or equity; in addition its specific character- istics and the dimension to which it refers have to be considered.
A taxonomy of social discrimination. Mummendey and Simon (1991) have offered a tax- onomy of social discrimination, which distinguishes two aspects: (a) the valence of resources (positive or negative) that are distributed between groups; and (b) the type of behavioral mode (direct/inflicting or indirect/withdrawing) by which this distribution is
established. In each of the resulting four cells, ingroup favoritism or outgroup antago- nism can be realized: One might opt for the own professional group getting a higher salary than another professional group, but expect the latter to pay more money when taxes are increased (direct and indirect discrimination in the positive domain); one might vote for another rather than the own home town as a place for an additional large garbage dump, but claim that the own community should be first in line when the township starts the re-naturation of areas that have been destroyed by the coal-mining industry (direct and indirect discrimination in the negative domain).
This variety of possibilities for differentiation and discrimination between groups is not reflected in the empirical work in this domain: Although everyday life provides many examples of a category-based infliction of burdens or costs in ways that the outgroup suffers relatively more than the ingroup, research has overwhelmingly dealt with differ- ential positive treatments.
Positive–negative asymmetry in social discrimination. Findings by Mummendey and col- laborators about the so-called positive–negative asymmetry in social discrimination (for a survey, see Mummendey & Otten, 1998; Otten & Mummendey, 2000) indicate that – at least for direct forms of discrimination – results obtained in the domain of positive resources should not be simply extrapolated to the negative area. The experiments mostly followed a typical minimal categorization procedure, but varied the valence of resources that were distributed between groups: On matrices like the ones in the original study by Tajfel and collaborators (1971) participants either allocated points that were allegedly to be transferred either into money, or into the duration of unpleasant noise or the number of unpleasant tasks (see Mummendey et al., 1992; Otten, Mummendey & Blanz, 1996).
In other studies, the manipulation of valence was realized by asking participants to eval- uate the novel ingroup and outgroup on either positive or negative trait dimensions (Blanz, Mummendey, & Otten, 1995). These and a number of subsequent experiments consistently showed that for both intergroup allocations and evaluations there were no favoritism effects if negative resources or evaluation dimensions were involved. In the pos- itive domain, however, the typical “mere categorization effect” was replicated: The (minimal) ingroup was significantly and positively differentiated from the respective out- group.
Integrating the empirical evidence from their studies, Mummendey and Otten (1998) propose the following explanation: Social categories that are minimal do not offer a legit- imate rationale for unequal intergroup treatment. Accordingly, in the negative domain a more elaborate cognitive processing (Otten, Mummendey, & Buhl, 1998) and a stronger concern about normative inhibitions (Blanz, Mummendey, & Otten, 1997; Otten &
Mummendey, 1999b) will raise the probability that the minimal categorization will be considered irrelevant. Thus, group members will re-categorize as a common ingroup (Gaertner, Dovidio, Anastasio, Bachman, & Rust, 1993) and refrain from differential intergroup treatment (Mummendey, Otten, Berger, & Kessler, in press). These findings were further corroborated in a recent study by Gardham and Brown (in press), demon- strating that subgroup and superordinate group identification were crucial for the effects of stimulus valence on intergroup treatment. Besides, this study is noteworthy as it manip- ulated not only valence but also behavioral mode (allocation, withdrawal). In fact, only
for instances of beneficiary decisions (i.e., allocating positive stimuli and withdrawing negative stimuli) was there significant ingroup favoritism. Finally, the importance of cat- egory salience in understanding the differential effects of stimulus valence on category salience was underlined in recent studies by Reynolds, Turner, and Haslam (2000), who argue that the positive–negative asymmetry might rely on valence-specific differences in the normative fit of the social categorization and the intergroup comparison dimension.
However, when testing the explanations for the potential explanations of the posi- tive–negative asymmetry effect, Mummendey and collaborators also documented that there are conditions, where ingroup favoritism and outgroup derogation were shown irre- spective of valence. Introducing so-called “aggravating conditions” (Blanz, Mummendey,
& Otten, 1995; Otten, Mummendey, & Blanz, 1996), such as inferior status and/or minority status, which increased the salience of the intergroup distinction and – possibly – elicited a threat towards positive social identity, resulted in significant favoritism effects in both valence conditions.
We can conclude from the above that negative forms of social discrimination can be demonstrated in the laboratory and in fairly reduced forms of intergroup settings.
However, when we turn from allocating minutes of unpleasant noise or samples of boring experimental tasks to much more dramatic negative treatments, then these effects necessi- tate a salient, subjectively meaningful social categorization that provides even greater legit- imization for a category-based intergroup treatment. In the following, we will analyze what such legitimizing rationales might look like, and how an allocating party might come up with the decision that it is appropriate or even inevitable to treat another group negatively.