2 Fundamental Orientations in Resource Allocations
3.2 Alternative Explanations of the MGP Discrimination Effect
identity within the intergroup setting. Results failed to support the BIM: autonomous individuals discriminated as much as interdependent participants on matrices such as FAV and MD. When the discriminatory behaviors of strong and weak identifiers were ana- lyzed, results showed that strong identifiers discriminated whereas weak identifiers did not. Results also showed that discrimination contributed to positive social identity only to the degree that participants identified strongly with their own group. Without ingroup identification, discrimination did not contribute to a more positive social identity. In an extension of the above study, Perreault and Bourhis (1998) showed that negatively autonomous participants, that is participants who were secretly told that they would per- sonally receive no money at all in the study, discriminated just as much as both interde- pendent and positively autonomous participants. In line with Dawes, van de Kragt, and Orbell (1990) these results suggest that ingroup identification can override concerns for personal self-interest. Taken together, these results show that ingroup identification (SIT) is a better explanation of discrimination in the MGP than self-interest and interdepend- ence as postulated within the BIM.
Although the above studies do not support the BIM they do not contradict Sherif et al.’s (1961) realistic conflict theory (RCT) conceptualization of interdependence and its complementarity with SIT (Turner & Bourhis, 1996). To the degree that individuals do identify with their own group, it is clear from Sherif ’s field studies that cooperative versus competitive group interdependence does have an impact on intergroup discrimination. In the Gagnon and Bourhis (1996) study, interdependence was manipulated at an individual level and not at the collectivelevel. A manipulation of negative autonomy at the collective level in which group members are denied valued resources by virtue of their group membership could be seen as illegitimate and unjust, thus triggering retaliatory behaviors such as discrimination against an outgroup seen as the cause of this inequity.
Uncertainty reduction
Using self-categorization theory as a conceptual backdrop, Hogg and Mullin (1999) pro- posed that the need for “uncertainty reduction” may be more fundamental than motiva- tion for a positive social identity (SIT) as an explanation of discrimination in the MGP.
The role of subjective “uncertainty reduction” in social identity processes is based on the premise that people have a basic need to feel certain about their social environment and their place within it. While subjective certainty contributes positively to the self by making existence meaningful and guiding behavior, uncertainty is aversive as it reduces mastery over one’s environment. It is proposed that individuals are more strongly motivated to reduce uncertainty in situations which are ambiguous than in those which are clear-cut, stable, and predictable.
According to Grieve and Hogg (1999), the classic minimal group paradigm (MGP) situation is high in subjective uncertainty due to the novelty of the ad hoc categorization (group X/Y), the undefined nature of relations between participants, and the “strange- ness” of the Tajfel matrices as a resource allocation task. They propose that participants within classic MGP conditions categorize themselves as members of their ad hoc ingroup mainly to reduce “subjective uncertainty” which in turn guides discriminatory behavior
and fosters ingroup identification. To better test their uncertainty reduction hypothesis, Hogg and Mullin (1999) manipulated the ambivalence of the MGP by creating experi- mental conditions which were less uncertain than the classic MGP situation. In three studies a “low uncertainty” condition was manipulated by giving participants a chance to practice with the Tajfel matrices for allocating points to uncategorized, individually num- bered recipients. Across the three studies results showed that categorized subjects in the usual “high uncertainty” MGP conditions did discriminate (composite score of FAV on P; FAV on MJP; MD), thus replicating the classic minimal group effect. However, the MGP effect was eliminated in the “low uncertainty” condition as subjects did not dis- criminate at all and tended to identify less with their own group than those in the classic
“high uncertainty” condition. With the MGP situation clarified in the “low uncertainty”
condition, individuals did not need to discriminate or to identify strongly with their ad hoc group as a way of reducing their subjective uncertainty. However, support for the uncertainty hypothesis was undermined by results showing that self-categorization and discrimination did not systematically reduce the pre-post uncertainty felt by participants in the experiments.
Though Hogg and Mullin (1999) did use pre-post measures of uncertainty reduction they only assessed ingroup identification aftersubjects completed the resource allocation task. Given that discrimination can increase both ingroup identification and positive social identity (Perreault & Bourhis, 1998), differential ingroup identification scores used as evidence of uncertainty reduction in the Hogg and Mullin (1999) studies may be the result of both uncertainty effects and the role of discrimination in increasing ingroup identification. Furthermore, the 10-item identification scale used in the three studies com- bined identification items with quality of social identity items thus confounding the dis- tinctive effects of discrimination on self-categorization (SCT) and on positive–negative social identity (SIT). As Rubin and Hewstone (1998) concluded, confounds within ingroup identification scales have contributed to the complexity of results exploring the link between collective self-esteem (quality of social identity) and discrimination. It is frustration with this very complexity that motivated some researchers to redirect explanations of discrimination from social identity needs (SIT) to uncertainty reduction needs.
It is clear that ongoing research must explore the distinctive and complementary roles of social identity and uncertainty reduction as key motives accounting for discrimination within the MGP. Moreover, within the usual “high uncertainty” condition of the classic MGP, Perreault and Bourhis (1999) did find that ethnocentric individuals were more likely to identify with their ad hoc ingroup (pre-post measures) than less ethnocentric persons (E and F scale measured a week before the MGP study proper). These results suggest that some individuals enter the MGP with a greater concern than others for iden- tifying with their ingroup and/or reducing uncertainty. Path analysis in the Perreault and Bourhis (1999) study also showed that unconfounded ingroup identification measured priorto resource allocations acted as a mediator variable predicting discrimination, while high–low ingroup identification manipulated as an independent variable also predicted discrimination. Thus, in line with SIT, identification with the ingroup category is a nec- essary antecedent condition for intergroup discrimination while unconfounded degree and quality of ingroup identity were each shown to increase following discrimination.
The positive–negative asymmetry effect
It is one thing to favor the ingroup through the distribution of positive allocations such as money, but quite another to discriminate by imposing negative outcomes such as burdens and punishments on outgroup members. In a series of MGP studies, Mum- mendey and Otten (1998) showed that categorization per se was not sufficient to trigger discriminatory behavior in the distribution of negative outcomes such as tedious tasks or unpleasant sounds. The failure to obtain the “minimal group discrimination effect” on negative outcome allocations poses a challenge to the generality of social identity theory (SIT) as an “explanation” of discrimination in the MGP.
Numerous studies conducted by Mummendey and colleagues have corroborated the
“positive–negative asymmetry effect” (PNAE) with both evaluative ratings (on positive versus negative traits) and on resource distributions (e.g., Tajfel matrices) using positive (money) versus negative allocations (unpleasant noise). Why is it that group members do not discriminate on negative outcome allocations in the MGP? Mummendey and Otten (1998) propose two classes of explanations to account for the PNAE within the MGP:
thenormativeand the cognitive. The normative perspective suggests that within the MGP, discrimination on negative outcome allocations is seen as even less socially acceptable than discrimination on positive outcome allocations. However, Mummendey and Otten (1998) concur with Billig and Tajfel (1973) in recognizing the circularity of post hoc nor- mative accounts for predicting parity and discriminatory behaviors.
The cognitive account draws from findings showing that negative outcome allocations elicit more careful information processing than positive ones. Mummendey and Otten (1998) propose that more accurate and deeper information processing for negative outcome allocations may make respondents more aware that there are no obvious grounds for treating ingroup and outgroup members differently within the MGP. From a self- categorization theory perspective (SCT) (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987), this awareness may shift self-categorizations from the group level categorization (us vs. them) to a supra-ordinate level: “all of us undergraduates” involved in the unpleas- ant task of distributing punishments to “fellow” undergraduates. From the recategoriza- tion framework proposed by Gaertner et al. (1993), the task of distributing negative outcome allocations may raise the salience of this supra-ordinate category (all of us under- graduates) thus reducing the pertinence of discrimination based on the experimenter- imposed “us–them” categories.
Otten et al. (1996) showed that low status and minority group members did dis- criminate on negative outcome allocations while subjects in identical “aggravating” con- ditions discriminated even more on positive outcome allocations. Invoking SIT, Otten et al. (1996) suggested that being in a low status or minority group position can threaten group members’ positive social identity. It is under such “aggravating” circumstances that discrimination on negative outcome allocations can be used to achieve a more secure and positive social identity. Unfortunately, Otten et al. (1996) did not measure feelings of group insecurity, perception of group threat, and pre-post quality of group identity to support their account of discrimination on negative outcomes in their studies. Recently, Blanz, Mummendey, and Otten (1998) found that degree of ingroup identification was
not related to valence asymmetry in their discrimination measures, a result leading to their rejection of social identity theory (SIT) as a heuristic explanation of the PNAE.
Unfortunately in this study, degree of ingroup identification was measured only after outcome allocations rather than both beforeandafter, suggesting that exclusion of SIT as a contributing explanation of the PNAE may be premature. Furthermore, from an SIT perspective, group members may consider that giving more “punishments” to the out- group than to the ingroup does not contribute to positive social identity. Unlike positive resources, negative outcome allocations may not constitute a valued or favorable com- parison dimension on which group members can “positively differentiate” from outgroup members.
Though current research is being conducted to replicate the positive–negative asym- metry effect in laboratories other than that of Mummendey and colleagues in Germany, a fundamental point remains: One can document numerous historical and current cases in which dominant group members systematically inflict extreme punishments on out- group others for an ideologically “good ingroup cause” while being self-aware of the socially unacceptable nature of their destructive actions (e.g., apartheid, ethnic cleansing, genocide).