Minimal Group Procedures - 1 Running Head: MINIMAL GROUP PROCEDURES An Improved (and Deception-Free) Minimal Group Induction Procedure

Two experiments compared alternative minimal group induction procedures involving (a) memorization of novel in-group names or (b) random categorization to (c) the traditional procedure based on ostensibly meaningful but deception-based categorization. Experiment 1 revealed comparable in-group attitude and identification effects across the three procedures on explicit measures, but significantly greater effects for the memorization procedure on implicit measures. Experiment 2 incorporated aspects of procedure (a) into (b) and (c). The modified procedures, while improved, were no more effective than the memorization procedure alone in creating novel in-group attitude and identification. In both experiments, implicit-explicit correlations were largest with the memorization procedure, which suggests greater validity. Conclusion: The memorization procedure can be used as a deception-free alternative to the traditional deception-based minimal group induction procedure. (124 words/854 characters) Minimal Group Procedures 3 An Improved (and Deception-Free) Minimal Group Induction Procedure Many years ago, a laboratory experiment produced curious findings: subjects who had been arbitrarily categorized into two novel groups subsequently displayed prejudicial attitudes and discriminatory behavior toward out-group members. Those initial findings from the minimal group paradigm (MGP; Tajfel, Billig, Bundy, & Flament, 1971; Rabbie & Horwitz, 1969) led to numerous theoretical advances in the study of groups, including social identity theory (Tajfel & Turner, 1986) and selfcategorization theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). Reports of MGP effects no longer provoke surprise. What perhaps should be surprising, however, is that more than 30 years since the creation of the MGP very little work has focused on refining the methods used to study intergroup bias and discrimination. Contemporary researchers use minimal group induction procedures that are remarkably unchanged from the procedure developed in the late 1960s. In its typical administration, subjects complete one of a number of tasks (e.g., estimating the number of dots briefly displayed on a screen or rating a series of paintings) and are subsequently informed that they responded to the initial task like one kind of person (e.g., a dot overestimator or the kind of person who prefers Klee paintings) and not another kind of person (e.g., a dot underestimator or the kind of person who prefers Kandinski paintings). Despite what subjects are led to believe, of course, the experimenter determines their assignment to the group. Later, subjects evaluate or allocate money to in-group and out-group members, with the typical result being that subjects significantly favor members of the groups to which they have been assigned. MGP Controversies, Revisions, and a Lingering Concern Tajfel and Turner (1986) viewed MGP effects as due to attempts to bolster one’s social identity. This perspective, central to Social Identity Theory, has remained an important explanation for MGP effects (Hewstone, Rubin, & Willis, 2002). Alternative Minimal Group Procedures 4 perspectives have emphasized explanations for why subjects discriminate that have little to do with image-bolstering, including expectations of reciprocity (e.g., Gaertner & Insko, 2000; Rabbie, Schot, & Visser, 1989) and the influence of norms (Hertel & Kerr, 2001). Researchers have also questioned the interpretation of intergroup bias effects shown in MGP studies, proposing alternative explanations that reflect cognitive consistency (Gramzow & Gaertner, 2005; Greenwald et al., 2002) rather than the motivational account of Social Identity Theory. These perspectives offer alternative explanations for MGP effects, but do not suggest alternative methods to produce those effects. One aspect of the MGP that has received challenge is the traditional assessment of intergroup discrimination, the Tajfel matrices. For this assessment, subjects must choose how to distribute money to in-group and out-group members from a set of allocation strategies. The choices vary in the absolute amount given to each group as well as the relative advantage given to one group over the other. The Tajfel matrices have been criticized for confounding the various allocation strategies (Brewer, 1979) and for providing limited or ambiguous response options (Bornstein et al., 1983a; 1983b). Bornstein and colleagues developed the Multiple Alternative Matrices (MAMs) to address these shortcomings. The MAMs provide a set of clear, unconfounded response options, including separate options that favor the in-group over the out-group (“max rel”) and maximize the absolute value for the in-group (“max own”). Bornstein et al. (1983a) did not find consistent evidence of in-group favoring allocations using the MAMs and so concluded that the typical MGP discrimination findings were produced by one or more artifacts in the Tajfel matrices. Gaertner & Insko (2001) subsequently showed that the MAMs do reliably measure intergroup discrimination provided that the allocation task does not invoke normative prohibitions against discrimination, as when the allocation task is framed as providing “payment,” rather than a “bonus.” Minimal Group Procedures 5 One methodological concern that has not yet been formally addressed is the MGP’s use of deception. Historically, across domains within social psychology, the use of deception appears regrettably common (Adair, Dushenko, & Lindsay, 1985). It is encouraging, however, that recent surveys of the literature suggest that the field’s use of deception is waning, particularly in the last few decades (Kimmel, 2001). In contrast to this trend, groups researchers continue to use the traditional MGP procedure for which subjects are misled regarding the true basis for their assignment to groups. This practice is all the more disconcerting considering that it is not intuitively obvious or, more importantly, empirically substantiated that use of deception is necessary to produce MGP effects. The present research directly examines this question in comparing traditional and novel, deception-free MGP procedures. Alternative Minimal Group Procedures A traditional alternative to the MGP deception-based procedure is a random categorization procedure, for which subjects are instructed that their assignment to groups is determined randomly. Does it make a difference which procedure is used? Although many studies have used the random categorization procedure, very few have directly compared both procedures. For example, Brewer and Silver (1978) had subjects complete the standard painting preference task and, in a similar condition, led subjects to believe that they had been assigned a group based on their similarities with others, whereas in an arbitrary condition, told subjects that their scores “were too similar to provide a basis for grouping, so they would have to be split into two groups randomly” (pp. 395-396). Note that deception was used in both conditions, but was arguably less central in the arbitrary condition. Brewer and Silver found that discriminatory allocations were similar in both conditions (see also Allen & Wilder, 1975). In contrast, Gaertner and Insko (2000) found that discrimination was reduced using a random procedure compared to a deception procedure. This result suggests that random Minimal Group Procedures 6 categorization may be less effective in instilling in subjects a meaningful self-group representation, which is characteristic of MGP effects. Research on the implicit partisanship effect (IP; Greenwald, Pickrell, & Farnham, 2002; Pinter & Greenwald, 2004) suggests a new alternative MGP method. Greenwald et al. had subjects study the names of four members of a novel group for 45 seconds. Their subjects subsequently completed assessments of implicit attitude and identification toward the studied group relative to an unstudied group using the Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998). They found that subjects (implicitly) significantly preferred and identified more with the studied group. Pinter and Greenwald (2004) conducted a series of studies to improve understanding of the IP effect. They found evidence for strong implicit preference of the studied group regardless of the type of relationship shared by the groups (i.e., competitive or cooperative) or the type of group itself (i.e., human or non-human). Further, Pinter and Greenwald discovered a condition in which the IP effect was limited: When subjects studied names of a novel group linked to a rival university, the effect of name-study was reduced. Might the procedure used to produce IP effects be used more generally as a MGP procedure? The IP and MGP procedures differ in a number of aspects. For one, MGP effects result from explicit categorization, whereas the IP effect presumably is a product of implicit categorization. For another, in contrast to the work on IP, MGP effects have been demonstrated with self-report measures, although increasingly researchers are using implicit measures in this domain as well (e.g., Ashburn-Nardo, Voils, and Monteith, 2001). Despite these differences, the important similarity between the MGP and IP effects is that they both demonstrate minimal group effects. This link suggests that a deception-free variant of the IP procedure might be adapted to study explicit