Download Minimal Group Paradigm: Factors Influencing Intergroup Discrimination in Arbitrary Groups and more Lecture notes Social Psychology in PDF only on Docsity!
Moderators of Intergroup Discrimination in the Minimal Group Paradigm: A Meta- Analysis Emily Pechar^1 Rachel Kranton 2 August 17 2017 Abstract: The minimal group paradigm (MGP) has been used as a method to demonstrate intergroup discrimination between arbitrary groups for the past 45 years. Recently, researchers have expanded the scope of MGP experiments to test what additional factors increase, decrease or moderate intergroup discrimination. However, a broad analysis of these findings is missing from the literature. Through a systematic review of minimal group experiments since 1970, we provide a greater understanding of what factors increase levels of intergroup discrimination in minimal groups. We find the strongest evidence for the impact of priming norms and increasing identity salience on intergroup discrimination. We also identify limitations of the current minimal group literature, including an overrepresentation of findings from white, middle-class, western samples and from experiments with student subjects. Building off of these findings, we offer recommendations on what future research on the minimal group paradigm should consider. Keywords: minimal group, identity, intergroup discrimination (^1) Duke University; PO Box 90328, Durham, NC 27708; emily.pechar@duke.edu (^2) Duke University; PO Box 90097, Durham, NC 27708; rachel.kranton@duke.edu
Introduction
One of the main tenets of social identity theory (Tajfel & Turner, 1979) is that maximizing the difference between one’s own group (ingroup) and another group (outgroup) increases positive social identity and self-esteem (Brewer, 1979; Oakes & Turner, 1980; Tajfel, Billig, Bundy, & Flament, 1971). Individuals tend to exhibit a disproportionate affinity towards members of their own group, and discriminate against members of other groups (known as intergroup discrimination). While such phenomena of prejudice and discrimination are evident in intergroup conflicts deep into history, it was only in the second half of the twentieth century that researchers began to experimentally test how social categorization per se can lead to intergroup discrimination. Early research on the drivers of intergroup discrimination had focused on the importance of social context and intergroup experiences. As one explanation of intergroup conflict, belief congruence theory emphasizes assumed differences in attitudes and beliefs between groups as the cause of prejudice and intergroup discrimination (Rokeach, 1960). Linville (1982) also draws on this theory, attributing intergroup discrimination to stereotypes attributed to the ingroup and outgroup as a result of greater familiarity with ingroup members. Drawing upon concrete differences between and conflict among groups, realistic group conflict theory (Campbell, 1965) points to a conflict in group goals or resources as a driver of intergroup discrimination. These perspectives all assume that intergroup discrimination is a symptom of intergroup relationships, resource conflicts, or stereotypes about the other group. However, beginning in the mid-twentieth century, researchers began to question whether these intergroup contextual factors were necessary to generate intergroup discrimination. Researchers began to ask: to what extent
MGP experiments show how intergroup discrimination is a tendency of human nature, even when the groups have no social meaning. Diehl (1990) identifies a number of key aspects of minimal group paradigm experiments that allow for this distinction between social and psychological drivers. First, the only difference between the two groups should be their membership in categories of a trivial nature. There should be no face-to-face interaction, and group membership of the other subjects should remain anonymous. Responses should be of no utilitarian value to the subjects themselves outside of the experiment or laboratory, yet they should be concrete rewards or punishments, not a qualitative evaluation of the individuals. Finally, the strategy of intergroup discrimination should be in competition with more rational or utilitarian approaches, and there should be no expectation of reciprocity or future interaction between the groups. By controlling for the social norms of preexisting attitudes, reciprocity and self-interest, MGP experiments demonstrate that humans have a tendency to discriminate against the outgroup and favor the ingroup in any situation. Tajfel’s 1970 experiment was designed to study the minimum necessary and sufficient conditions for intergroup discrimination. Since then, researchers have sought to expand on Tajfel’s findings by testing what factors and situational contexts contribute to or increase intergroup discrimination between minimal groups. Numerous studies have added in “enhancements” to the initial MGP design to determine how they contribute to or moderate intergroup discrimination. These enhancements have been diverse, from analyses looking at how different types of group categorizations can increase the threat to self-esteem and increase intergroup discrimination (Abrams & Hogg, 1988; Diehl, 1990), to testing the role of self-esteem (Lemyre & Smith, 1985) and attitude congruence (Allen & Wilder, 1979), among others. While the literature on MGP enhancements has grown, it is difficult to broadly understand what we
know about the moderators of intergroup discrimination between minimal groups. Additionally, some recent work has called into question the universality of intergroup discrimination in minimal groups, finding evidence that MGP studies do not consistently result in discrimination among all individuals (Kranton, Pease, Sanders, & Huettel, 2016). These findings call for a broad review of studies to better understand the contexts in which intergroup discrimination is most likely to arise. To clarify the state of the MGP literature, this study identifies and analyzes the level of consensus on what factors increase intergroup discrimination. Through a systematic review of minimal group studies from the past 45 years, we identify the contexts, traits and enhancements that lead to increased intergroup discrimination in minimal group situations. Following the guidance on MGP studies laid out by Diehl (1990), we identify 84 minimal group studies in the social science literature. We then categorize them to draw conclusions about what enhancements increase levels of intergroup discrimination, provide descriptive information about the temporal and geographic trends in the literature, and offer an analysis of gaps that should be addressed in future MGP studies.
Methods
While hundreds of studies have expanded on Tajfel’s original MGP experiments, a clear understanding of the factors that exacerbate or diminish intergroup discrimination does not yet exist. This meta-analysis systematically evaluates minimal group paradigm studies from 1970- 2015 to identify trends in findings (which experimental enhancements and subject traits increase intergroup discrimination) as well as publication characteristics to provide a broad state of the
- Geographic location of lead author
- Year
- Title
- Journal title
- Journal impact factor (as of 2016)^3
- Sample size
- Sample type (children, students, or adults, and any further details about the sample that were provided)
- Whether the study tested an experimental enhancement or a personal trait
- What that enhancement or trait was
- An abstract-level summary Analytical Strategy The meta-analysis involved both qualitative and quantitative analyses. The first analysis involved qualitative categorizations of the enhancements and traits that each study tested as moderators of intergroup discrimination. The intention of this analysis was to gain a more concise understanding of what features researchers have evaluated as moderators of discrimination between minimal groups before performing a quantitative analysis of the effectiveness of each type of enhancement in increasing intergroup discrimination. After reviewing each article, we grouped studies into ten groups based on similarities in the enhancements used in each.^4 The categories included:
- Norms (priming social norms such as loyalty or competitiveness)
- Group Status (high or low status of outgroup compared to the ingroup)
- Identity Salience (how strongly subjects identify with their group)
- Outcome Type (positive or negative valence of the outcomes being allocated) (^3) Thompson Reuters 2015 Journal Citation Reports Impact Factors (^4) Studies were categorized under multiple groups if they tested several types of enhancements.
- Personality (traits of the subjects’ personality)
- Group Size (whether the outgroup is smaller or larger)
- Gender & Age (of the subjects)
- Mood (emotional state of mind)
- Group Categorization (number of groups, their interdependence, and manner of categorization)
- Miscellaneous (including physiological traits, culture, and knowledge). The qualitative review involved a close reading of the studies in each category and the summarization of their findings in narrative form. From this we draw category-level conclusions about the moderating impact that each enhancement has on levels of intergroup discrimination After compiling and summarizing conclusions for the categories, we performed a quantitative analysis to measure the level of consensus among studies about whether the enhancements summarized in each category actually increased intergroup discrimination. We coded each study within a category to capture whether that enhancement or trait increased discrimination at a statistically significant level using a “ 0 - .5- 1 ” coding system. In this system, a study was assigned a “0” if the enhancement/trait was not found to increase discrimination between minimal groups, a “.5” if the evidence was mixed, and a “1” if it was definitively found to increase discrimination. After coding the studies, we calculated an average of all the individual study codes in each category to represent the level of consensus in the literature about the impact of each enhancement on intergroup discrimination (the category consensus score). We also reported the enhancement that the studies measured as a moderator of intergroup discrimination, the average impact factor of the studies in the category (to indicate prominence of the studies in the literature), and the median publication year (with minimum and maximum date ranges).
Other studies focused on the effect of priming the subject to feel that the outgroup is immoral (Abrams et al., 2008; Chen & Li, 2009; Hetherington, Hendrickson, & Koenig,
- or unfair (Diehl, 1990; Jetten, Spears, & R, 1996). Establishing the outgroup as immoral increases the psychological distance between how the subject perceives him or herself and the outgroup, and may also elicit a desire to “punish” immoral outgroup members. Finally, several studies looked not at priming a particular norm, but allowing groups to coalesce around some ingroup norm of behavior. This generally occurs when the experiment allows for communication between ingroup members before making allocation decisions. The expectation of most of these studies is that establishing a group norm will result in increased discrimination. However, none of the studies evaluating this norm found evidence of increased intergroup discrimination (Amiot & Bourhis, 2005; Billig, 1973; Smith & Postmes, 2009).
2. Group Status Expanding on the original MGP studies where groups were perceptually equal, many studies have looked at how intergroup discrimination changes when the groups are of different statuses. These studies define status in a number of ways, encompassing wealth, size, gratification/deprivation, and merit, but in each the group status is experimentally induced. Subjects are told that the groups are of unequal status, and that they are in either a high, low, or moderate status group. Many of the studies found that subjects in low status groups show more intergroup discrimination than high status groups, which aligns with expectations from social identity theory (SIT). According to SIT, humans exhibit
intergroup discrimination to increase self-esteem, which would be an expected objective of lower status groups. However, findings of the studies in this category were inconsistent. While several studies found that low status groups discriminated more than high status groups (Caricati & Monacelli, 2010; Harvey & Bourhis, 2012; Moscatelli, Albarello, Prati, & Rubini, 2014; Otten, Mummendey, & Blanz, 1996; Platow et al., 1997), others found mixed results (Amiot & Bourhis, 2005; Dobbs & Crano, 2001; Harvey & Bourhis, 2013; Reichl, 1997; Rubin, Badea, & Jetten, 2014; Rubini, Moscatelli, Albarello, & Palmonari, 2007). Some also found no evidence of increased discrimination among low status groups (Rothgerber & Worchel, 1997), or that low status groups discriminate less (Moscatelli et al., 2014; Sachdev & Bourhis, 1991). A few additional moderators identified in these studies may offer keys to understanding when status matters, and when it does not. Caricati (2010) found that when resources being allocated are perceived as unlimited, this reduces bias in low status groups. Amiot & Bourhis (2005) found that low-status groups are more discriminatory until the group members are allowed to communicate and establish ingroup norms, after which the high and low status groups discriminate equally. Dobbs & Crano (2001) provided evidence that the salience of intergroup differences may matter: when subjects were required to provide a justification for their allocations, discrimination decreased in high status groups, but increased in low status groups. Harvey & Bourhis (2013) built on the norms of fairness by demonstrating that the merit of group status matters. If group status is obviously random or undeserving, there is less discrimination (particularly among the high status groups). Similarly, Reichl (1997) showed increased levels of discrimination between high and low
not affect level of intergroup discrimination, when combined with a mood manipulation, those in a sad mood increased discrimination when group identity is salient. Linking to the group status literature, Leonardelli & Brewer (2001) found that majority group members only discriminate against an outgroup when their group identity is made more salient.
4. Outcome Type Building off the Positive-Negative Asymmetry Effect (PNAE) previously established in social discrimination research (see Buhl ( 199 9)), several studies tested how the valence and type of outcomes allocated increases or decreases levels of intergroup discrimination. The PNAE states that individuals discriminate more when allocating positive outcomes than negative ones. This finding was supported by three studies in this sample (Gardham & Brown, 2001; Hodson, Dovidio, & Esses, 2003; Jost & Azzi, 1996), although some found no evidence that outcome valence increases intergroup discrimination, or that it does so only under certain contexts (Amiot & Bourhis, 2003, 2005). Specifically, Amiot & Buorhis (2005) found that the PNAE only emerges after groups meet to form a group behavior norm. However, Smith & Postmes (2009) found the opposite – that the PNAE exists until group communication occurs. Otten et al (1996) found support for the PNAE unless the individual is a member of a minority or inferior group, in which case discrimination occurs equally for both types of outcomes. Researchers have also evaluated how outcome type and relevance increases discrimination. Gaertner & Insko (2000) found that intergroup discrimination is particularly heightened when allocating bonus payments as opposed to monetary wage payments, providing support for the idea that subjects are more likely to discriminate on unexpected or “additional” payments. Similarly, Jost & Azzi (1996) found that subjects are less
discriminatory with monetary payments than with less tangible resources, such as calculator time to help solve math problems. Finally, as discussed briefly above, some subjects show a tendency to discriminate more on outcomes that are highly relevant to either the group status or the activity involved (Forgas & Fiedler, 1996; Hartstone & Augoustinos, 1995; Reichl, 1997).
5. Personality Beyond context-specific enhancements, researchers have also sought to determine if there are aspects of an individual’s personality that make them more or less likely to show intergroup discrimination in the MGP. While the results of these studies are mixed, there is evidence from between-subject experiments that a few personality traits consistently seem to increase discrimination in the MGP. Most prominent is low self-esteem, which has been shown in a number of studies to increase intergroup discrimination (Falk, Heine, & Takemura, 2014; Hogg & Sunderland, 1991; Hunter et al., 2005; Petersen & Blank, 2003). In these studies, self-esteem is measured prior to the MGP procedure, and then discrimination levels are compared between subjects with different pre-experimental self- esteem levels. This is again supported by one of the main tenets of social identity theory, which asserts that intergroup discrimination is done to increase self-esteem. Other studies have shown that subjects with an independent self-construal (thinking of oneself first as an individual) show discrimination primarily when self-esteem is threatened, while interdependent self-construal subjects (those that think of themselves first as members of society) show more discrimination during instances of low threat to self-esteem (Nakashima, Isobe, & Ura, 2008). Prosocial and antisocial tendencies also contribute to discrimination, with prosocial (less competitive) subjects showing less discrimination than
7. Gender & Age The studies in this category are unique. While most MGP studies find no difference in responses between males and females and therefore do not include them further in their analysis or report discrimination levels by subgroup, several studies did find significantly different levels of discrimination between males and females. For the most part, these studies found that males discriminated more than females. This trend seems to be strongest among younger children (Benozio & Diesendruck, 2015; Hartstone & Augoustinos, 1995). Specifically, Benozio & Diesendruck (2015) found that the increased discrimination among boys (compared to girls) disappeared in samples with older boys and girls. Yuki & Yokota (2009) found some evidence of increased discrimination among adult men, but Gaertner & Insko (2000) showed the opposite – that women discriminate more than men. Because the evidence is mixed and most other studies find no differences in discrimination levels between men and women, we cannot confidently find any consistent trends of how gender may increase discrimination. Similarly, most MGP studies are done with a homogenous sample (overwhelmingly with university students). Of the few that do test different age groups, most find that younger children show more intergroup discrimination, offering some insight that perhaps the reduction in intergroup discrimination is learned. Benozio & Diesendruck (2015) performed their experiments on 3 and 6 year olds, and find that discrimination increases among the 3 year olds. While Plotner et al (2015) did not find a statistically significant outgroup bias in resource allocation, they did find that evidence that prosocial behavior towards the ingroup increases among 5 year olds but not 3 year olds. 8. Mood
Researchers have also investigated the relationship between emotions, or mood, and intergroup discrimination. In three different experiments, Forgas & Fielder (1996) tested how priming a positive or negative mood increases discrimination. While they initially found that subjects were more discriminatory when in a positive mood, they also reported that this effect is qualified based on personal relevance of the group. When personal relevance is low, a positive mood resulted in more discrimination. When personal relevance is high, a negative mood resulted in more discrimination. Others have looked at how an uncertain mood can impact discrimination. Intergroup discrimination can increase self-esteem, but emphasizing a group identity can also help to reduce feelings of uncertainty (Hogg, 2000). Based off of this finding, several studies have looked more closely at the role that confusion or an uncertain state of mind plays in intergroup discrimination. Grieve & Hogg (1999) and Mullin & Hogg (1998) demonstrated that when subjects are placed into contexts designed to increase feelings of uncertainty (such as watching a confusing short film or not being able to solve an ambiguous puzzle), their resulting levels of intergroup discrimination are significantly higher than those in control groups. However, Hodson & Sorrentino (2001) showed that this effect is mediated by personality. In their study, they found no difference in discrimination between subjects in high uncertainty and low uncertainty contexts in general with the sample as a whole, but did find evidence that high uncertainty increases discrimination among certainty-oriented people.
9. Group Categorization Three elements of group categorization were tested in the studies reviewed to determine how features of the MGP methodology might increase or decrease
Finally, there were several studies in the sample that tested enhancements or traits that were unique to any other studies. Some research has been done on how physiological traits inform levels of intergroup discrimination. Sahdra et al (2015) found that individuals with more heart-rate variability show greater levels of discrimination. Volz et al (2009) found that social and personal identity processes draw on the same cerebral correlates using an fMRI during a MGP experiment. Falk et al (2014) focused not on physiological but on cultural traits to explain differences in discrimination. They found that Americans show higher levels of discrimination in MGP experiments than do Japanese subjects, partially mediated by self-esteem. Lastly, Yamagishi & Mifune (2008) tested how knowledge of the identities of each group moderates discrimination. They found that ingroup bias is more commonly found when both the allocator and the recipient know each other’s’ group identities, but disappears when the allocator knows the identities but the subject does not. Quantitative Category Consensus In addition to the qualitative summaries of each enhancement category, we also performed a quantitative analysis of the level of consensus among the studies on the moderating role of each enhancement on intergroup discrimination. Table 1 presents the results of this analysis. The table includes the number of studies and level of consensus for each category, as well as descriptive features of the studies in each category, including the average journal impact score and range and median publication dates. The categories are listed in each row by number of studies in the category, followed by an indicator symbol that reflects the category’s
prominence and consensus level in the literature.^5 The indicator circle reflects both the number of studies in the category, and the consensus between the studies about whether the enhancement increases discrimination. The size of the circle reflects the number of studies in each category^6 (the larger the circle, the more studies in the category), and the darkness gradient reflects the level of consensus among the studies about whether that enhancement increases discrimination (the darker the gradient, the higher the consensus score for that category). [Table 1 about here] The most frequent enhancement in the study sample norms priming, comprising studies that primed norms such as loyalty, equality or competitiveness among subjects prior to performing the minimal group experiment. Consensus about the effectiveness of norm priming in increasing intergroup discrimination was moderate ( 0 .63 out of 1.0), signifying that about two thirds of the studies that tested this enhancement found that this tactic increased intergroup discrimination. This category of enhancements is also among the most recent MGP studies, with a median publication year of 2008 (although also the widest range, with the first norms study published in 1973). Group status also included a fairly high number of studies (17), however the consensus score was less than 50% (0.47), indicating a great deal of disagreement among studies about whether differences in the status of groups can increase intergroup discrimination. Specifically, the findings of studies about how high- and low-status groups discriminate differently often conflicted with each other. (^5) Miscellaneous category not included in quantitative analysis or Table 1 (^6) Studies were only included in the quantitative analysis if there was an enhancement being tested. Basic replications of the MGP were not categorized.
