The psychological primacy of the ingroup seems axiomatic in research and theory on intergroup relations. As Brewer (1999) notes, “A cursory review of forty years of social psychological research on intergroup relations suggests that Allport (1954) was right in assigning psychological primacy to the processes of ingroup formation and attachment over attitudes toward outgroups (pp. 441–442).” Emotional connection to the group is a key component of social identity (Hogg, 2003), and positive affect for an ingroup is a direct function of identification with the group (Correll & Park, 2005); not surprisingly then, ingroup “love” is deemed more fundamental than outgroup “hate.” Similarly, intergroup bias is considerably more likely to involve ingroup favoritism than outgroup derogation (Brewer, 1979, 1999; Brown, 2000; Hodson, Dovidio, & Esses, 2003; Hogg, 2003; Mummendey et al., 1992; Mummendey & Wenzel, 1999; Park & Judd, 2005). Using a novel game paradigm designed to specifically distinguish between ingroup “love” and outgroup “hate,” Halevy and colleagues (Halevy, Bornstein, & Sagiv, 2008; Halevy, Weisel, & Bornstein, 2012) found that in both single-shot trials and repeated interactions contributions were made almost exclusively to the pool that benefited the ingroup (at a personal cost) rather than to the pool that also harmed the outgroup. These studies provide further support for the claim that ingroup “love” is a stronger motivation than outgroup “hate” for group members.

As noted by Correll and Park (2005), ingroups are a resource and provide many psychological benefits, but the psychological value of ingroups does not necessitate that we derogate outgroups; that is, ingroup “love” need not lead to outgroup “hate.” Nevertheless, there may be at least one category of group membership that necessarily entails outgroup derogation, and that is the case of morality-based groups.

Reflecting our social interdependence, morality entails standards and rules developed to facilitate group living and social inclusion (De Waal, 1996; Haidt, 2007; Haidt & Joseph, 2004; Janoff-Bulman, Sheikh, & Hepp, 2009; Krebs, 2008). Interest in the psychology of morality has mushroomed in the past decade, broadening our understanding to include a central role for affect and intuition beyond moral reasoning (e.g., Greene, Sommerville, Nystrom, Darley, & Cohen, 2001; Haidt, 2001, 2008). Morality is not simply about good and bad, which may be applied to matters of taste and preference, but rather about right and wrong. Importantly, judgments of harm are fundamental to an understanding of morality; that is, “morality is understood through the lens of harm” (Gray, Young, & Waytz, 2012, p. 109). When asked to think of an immoral action, people spontaneously report harmful acts, typically involving an intentional agent and a suffering victim (Gray et al., 2012). Also, for many years researchers have distinguished between moral transgressions and conventional transgressions based on the former’s association with harm (Nucci, 1982, 2001; Turiel, 1983). And although recent work by Haidt and colleagues (Haidt, 2007, 2008; Haidt & Joseph, 2004; Haidt & Graham, 2007) has expanded the moral domain to include seemingly non-harm-related concerns with authority, purity, and loyalty, Gray, Young, & Waytz (2012) have demonstrated that transgressions in these domains implicitly activate considerations of harm, and actual victims (implying harm) are identified in these cases as well. Support for the harm-based characterization of moral convictions is also evident in work on moralization, for attitudes become moralized through their association with harm-doing, and greater perceptions of harm lead to stronger moral convictions (see, e.g., Rozin, 1999).

Based on judgments of harm, moral convictions are absolute and perceived as universal (Hare, 1981; Kant, 1786/1947; also see Haidt, Rosenberg, & Hom, 2003); they are “experienced as facts about the world” (Skitka, Bauman, & Sargis, 2005, p. 896.). We may recognize that others hold a different moral position, but we regard them as “wrong” and believe they should share our view of morality. In our evaluation of (im)moral positions and the people who hold them, there is no continuum of “rightness,” but rather two possibilities—moral and immoral (see, e.g., Gray & Wegner, 2009; Horberg, Oveis, Keltner, & Cohen, 2009; Krebs, 2008; Skitka, et al., 2005). These are oppositional in nature: if my position is right, your (different) position is wrong.

Researchers have established the uniqueness of attitudes based on morality—moral convictions—as distinct from equally strong attitudes not rooted in morality (Mullen & Skitka, 2006; Skitka et al., 2005; Skitka & Houston, 2001). Attitudes rooted in morality have considerable motivational force or action potential in directing our behavior and are typically accompanied by strong emotions (Skitka et al., 2005; also see Haidt, 2001, 2007; Rozin, Lowery, Imada, & Haidt, 1999). We react to those whose moral convictions differ from our own with intolerance, a strong desire for social and physical distance, and low levels of cooperativeness and goodwill (Skitka et al., 2005; also see Haidt et al., 2003, Mullen & Skitka, 2006). Disagreements in the domain of morality not only lead to a deceased ability to resolve differences but also to a loss of concern with procedural safeguards (e.g., due process of law) when pursuing ends consistent with one’s own moral position (Skitka et al., 2005; Skitka & Houston, 2001; Skitka & Mullen, 2002).

Given the uniqueness of moral convictions, we might assume that groups and related social identities based on these beliefs might display some distinct characteristics as well. Many researchers have emphasized the importance of beliefs for social identity (for a review, see Correll & Park, 2005), and shared beliefs can be the basis for social group formation (Bar-Tal, 1998). Moral convictions, too, can be the basis of social identities and group memberships. For example, attitudes about abortion are the basis for labels denoting distinct social groups (i.e., “pro-choice” and “pro-life”); these are morality-based groups. Based in shared beliefs involving judgments of harm, morality-based groups advance a position on a moral conviction; that is, the groups bind together those individuals who share a conviction that a particular action or behavior is either right (i.e., moral) or wrong (i.e., immoral). Given opposing positions, morality-based groups are typically characterized by the dichotomous categorization of two groups in opposition to one another (e.g., pro-life vs. pro-choice on abortion).

We would expect reactions toward morally dissimilar others (e.g., intolerance, desire for physical distance), studied at the interpersonal level, to be evident in reactions toward groups and group members with opposing moral views as well. We not only hold different views on matters of fundamental importance but opposing views that define our group memberships. In fact, in morality-based groups, ingroup identification seems fundamentally based on the existence of an outgroup; social identity itself would lose all meaning without the existence of the outgroup. Here, the ingroup and outgroup simultaneously define each other because of the oppositional nature of morality. Morality-based social identities are likely composed of both affirmational and negational components (see Zhong, Phillips, Leonardelli, & Galinsky, 2008, on negational categorization). Because the beliefs of the ingroup and the outgroup are oppositional, to identify with the ingroup is also to identify as “not them.” The identity one has with a morality-based group is thus likely to have components of both ingroup “love” and outgroup “hate.”

Not all moral issues will be the basis for morality-based groups. Where there is social consensus, groups are unlikely to form. Thus, the belief that incest is wrong is a strongly held moral view with consensus in our society (Haidt, 2001); there is no salient moral social identity based on this issue, because there is no outgroup representing a strong opposing view. Morality-based groups involve oppositional outgroups, and identification with the ingroup, we suggest, is based largely on perceived threat and a desire to distance oneself from the outgroup. Whereas my group’s beliefs are moral and right, the outgroup’s beliefs are immoral and wrong. Divergent moral beliefs implicate a deep-seated difference regarding the very rules we live by—and given the fundamental role of perceived harm in morality (and thus groups based on moral convictions), threat is particularly likely to characterize intergroup perceptions and promote negative outgroup perceptions (see, e.g., Riek, Mania, and Gaertner, 2006; Stephan & Renfro, 2002; Stephan & Stephan, 2000).

Even in the case where the group dyad is not as clear as that of abortion, as in the case of environmental groups such as Greenpeace, the group members are bound together by the common perception of harm by those who do not protect the environment (e.g., companies who transport and sell fossil fuels). Conversely, those who are on the other side will perceive the pro-environmental groups as a menace causing harm to the economy. Therefore, psychological threat will arise for both groups if the dyad is truly morality-based. Because of perceived harm, and also the intent to harm, threat will necessarily be an integral part of the moral convictions that become the basis for morality-based social identity. In the context of morality-based groups, each group perceives the other as the harm-doer.

When social group memberships are defined by moral conviction, the unique nature of morality suggests that outgroup negativity will play a central role. Interestingly, past research on intergroup relations has been inconsistent in outlining when identity concerns depend on or motivate negative outgroup attitudes (see Brown, 2000), but the relationship between positive and negative emotional reactions to ingroups and outgroups has consistently received less attention than intergroup bias itself. We suggest that moral social identities are a special classification of social group memberships in which outgroup “hate” naturally co-exists with ingroup “love.” For most groups, social identity is based solely on the ingroup and what it offers its members; here ingroup “love” is paramount. In the case of morality-based groups, social identity implicates opposition to the outgroup, and threat from the opposing group leads to outgroup “hate.” Somewhat attenuated ingroup positivity is likely to result when outgroup negativity also becomes a basis for group identification.

The three studies that follow explored both ingroup positivity and outgroup negativity in morality-based and non-morality-based groups. We expected outgroup negativity to play a more powerful role for the former. In particular, we hypothesized that compared to non-morality-based groups, morality-based groups would evidence lower ingroup positivity and greater outgroup negativity and perceived threat; and further, that ingroup positivity and outgroup negativity would be interdependent.

Study 1

For our morality-based group, we chose a familiar hot-button moral issue in the US, that of abortion, to help ensure the moral basis of the group identities; all participants had strong convictions (“right” or “wrong”) regarding legal abortion. For our non-morality-based groups, we sought a population that would also have strong convictions, but not in the moral domain. We chose the sports domain–specifically the Red Sox-Yankees rivalry (the current studies were conducted in Massachusetts). These sports fans seemed like a particularly appropriate comparison, for we believed that emotional intensity, which is strongly associated with moral attitudes, would also characterize our baseball team fans.

Method

Participants

A total of 139 undergraduate students participated in the study for course credit. There were 82 participants (64 females) in the morality-based (MB) group and 57 participants (42 females) in the non-morality-based (N-MB) group. The participants were recruited based on their responses to a prescreening questionnaire administered at the beginning of the semester. Participants recruited for the MB group indicated strong support of legal abortion (i.e., 8 or 9 on a 9-point scale) and strong opposition to making abortion illegal (i.e., 1 or 2). They also indicated that the issue of abortion was very important (i.e., 8 or 9 on a 9-point scale). Participants for the N-MB group were only recruited if they strongly approved of the Boston Red Sox (i.e., 8 or 9 on a 9-point scale), strongly disapproved (i.e., 1 or 2) of the New York Yankees, and indicated that the Red Sox team was highly important to them (i.e., 8 or 9 on a 9-point scale).

Materials and Procedure

Participants completed a questionnaire that assessed group-based emotions and perceived threat for both the ingroup and the outgroup (counterbalanced). On 9-point scales (0 = “Not at all”; 8 = “Extremely”), participants indicated the extent to which each group made them feel angry, happy, disgusted, respectful, fearful, positive, anxious, resentful, proud, hurt, sad, and negative. Ten items taken from Cottrell and Neuberg (2005) assessed threat using 9-point scales (−4 = “Strongly Disagree”; 4 = “Strongly Agree”). The items measured threats to American society, violations of trust, threats to personal values, threats to freedoms and rights, threats to norms of fairness and justice, threats to American values, threats to economic opportunities, threats to physical health, threats to physical safety, and threats to the overall functioning of American society. Examples of these items (with phrasing representing the distinct target groups) include: “People who are pro-life are a threat to my values,” “People who are pro-choice threaten my personal freedoms and rights,” “People who are fans of the Boston Red Sox threaten economic opportunities of others,” and “People who are fans of the New York Yankees are a threat to the physical safety of others in society.”

Results and Discussion

Aggregate measures of positive ingroup emotions (α = .92) and negative outgroup emotions (α = .90) were created by averaging across items. Negative ingroup emotions and positive outgroup emotions produced no meaningful variability for either group type and were highly skewed variables; therefore, these were not subjected to any analyses and will not be discussed further. The two emotion measures representing ingroup positivity and outgroup negativity were subjected to a 2 × 2 mixed ANOVA with the group type as a between subject factor. A significant interaction emerged between group type and emotion type, F(1,137) = 49.27, p < .001, \(\eta_{p^{2}} = .26\) (see Fig. 1). As hypothesized, participants felt more positive about the N-MB ingroup than the MB ingroup, t(137) = 4.61, p < .001, d = 0.80. Conversely, the MB outgroup elicited more negativity than N-MB outgroup, t(137) = 3.76, p < .001, d = 0.62. Participant gender did not moderate any of these effects, and will not be discussed further.Footnote 1

Fig. 1
figure 1

Positive ingroup emotions and negative outgroup emotions as a function of group type

Full size image

As with negative ingroup emotions, there was very little meaningful variability and extremely high skew for threat items directed at the ingroup. Therefore, only threat items directed at the outgroup (α = .95) were analyzed. As expected, MB outgroups were perceived as more threatening than N-MB outgroups (Ms = 0.47 vs. −2.45), t(137) = 9.49, p < .001, d = 1.64. To test whether threat is a mediator of the difference in outgroup negativity, a regression analysis was conducted with the group type (MB = 1, N-MB = 0) and outgroup threat as predictors of negative outgroup emotions. Whereas a significant difference between the groups (b = 1.13) was previously obtained without controlling for threat, the difference was entirely accounted for when controlling for threat, b = −0.39, SE = 0.35, p = .27. Threat was a significant predictor of negative outgroup emotions when controlling for the group type, b = .52, SE = 0.07, p < .001. The indirect effect was computed using bias-corrected (BC) bootstrapping with 5,000 resamples (see Preacher & Hayes, 2008). We concluded that threat significantly mediated the relationship between the group type and outgroup negativity, 95 % CI [1.02, 2.18].

Study 2

The purpose of Study 2 was primarily to extend the findings of Study 1 by ruling out group status as an alternative explanation. Because Study 1 involved only participants from the regionally normative groups, we recruited participants from each of the four target groups from Study 1. Also, we measured identification in Study 2 instead of using the strength of identification to recruit participants. This allowed us to test the relationships between the evaluations of the ingroup and the outgroup with identification for each group type.

Method

Participants

Participants were 134 (99 females, 30 males, 5 unknown) undergraduates who participated for course credit. Majority and minority status of the groups were validated by prescreen data used for Study 1 (N = 1,709), which indicated that pro-choice participants outnumbered pro-life participants by a 4:1 ratio in this population, and Red Sox fans outnumbered Yankees fans by a 7:1 ratio. Participants were included in the MB group if they responded with a −3 or −4 (opposed to legal abortion; n = 27) or +3 or +4 (in favor of legal abortion; n = 48) on the measure of attitudes toward abortion. Participants were included in the N-MB groups if they indicated that they strongly approved (+3 or +4) of either the Red Sox (n = 36) or the Yankees (n = 23). Participants were not included if they were above the midpoint on both the scales.

Measures

Participants completed the same threat and emotion measures used in Study 1, counterbalanced for ingroup versus outgroup. The primary variables of interest were outgroup threat (α = .96), positive ingroup emotions (α = .92), and negative outgroup emotions (α = .91). Feeling thermometers were used for global evaluations of the ingroup and outgroup. These scales indicated how warm they felt about each group (0 = “Very Cold”; 100 = “Very Warm”). Participants also indicated the extent to which the ingroup was important to them on a 9-point scale (1 = “Not at all important”; 9 = “Very important”).

Results and Discussion

Group Status

There were no differences between MB and N-MB participants’ ratings of the importance of the ingroup, t(132) = 1.17, p = .24, d = .20. Using a series of 2 × 2 between-subject ANOVAs, we replicated the findings from Study 1. The pro-choice and pro-life (MB) groups (M = 3.23, SD = 2.09) both had significantly higher scores on outgroup negativity than the N-MB groups (M = 1.82, SD = 1.44), F(1,130) = 24.41, p < .001, \(\eta_{p^{2}} = .16\). The interaction between status and group type was also significant, F(1,130) = 4.08, p = .046, \(\eta_{p^{2}} = .03\). Instead of ruling out the findings from Study 1, this interaction suggests the level of outgroup negativity is even stronger for the minority (i.e., pro-life) group. Moreover, the difference between the MB and N-MB groups when separated by status indicated that the main effect of group type was not being driven by the minority group alone. As with Study 1, pro-choice group members (M = 2.77) demonstrated more negative outgroup emotions than did fans of the Red Sox (M = 1.83), t(82) = 2.32, p = .023, d = 0.51. Pro-life group members (M = 4.03) also felt more negativity toward the outgroup than did fans of the Yankees (M = 1.79), t(48) = 4.61, p < .001, d = 1.31.

Members of MB groups (M = −0.02, SD = 2.14) also perceived significantly more threat than members of N-MB groups (M = −2.30, SD = 1.79), F(1,130) = 40.31, p < .001, \(\eta_{p^{2}} = .24\), and there was no interaction or main effect of status, Fs < 1. The effect was significant (ps < .001) and similarly large for both the majority (d = 1.10) and the minority (d = 1.19) groups. We again tested whether threat mediated the relationship between the group type and outgroup negativity, and we found that greater threat was associated with negative outgroup emotions while controlling for the group type, b = 0.55, SE = 0.06, p < .001. The difference between the groups on outgroup negativity (b = 1.41) was also entirely accounted for by threat, as MB groups were no longer different from N-MB groups after including threat into the model, b = 0.15, SE = 0.30, p = .61. As in Study 1, the indirect effect was significant using BC bootstrapping with 5,000 resamples, 95 % CI [0.81, 1.82].Footnote 2

Finally, the N-MB groups (M = 6.81, SD = 1.55) both had higher ingroup positivity scores than the MB groups (M = 5.41, SD = 2.25), F(1,130) = 14.23, p < .001, \(\eta_{p^{2}} = .10\). The main effect of status was not significant, F(1,130) = 1.18, p = .28, \(\eta_{p^{2}} < .01\), nor was the interaction significant, F < 1. Testing the majority and minority groups separately revealed the differences between MB and N-MB groups were significant for both the majority and the minority groups. The pro-choice group felt less positivity toward the ingroup than did Red Sox fans, t(82) = 3.51, p < .001, d = .77. Similarly, the pro-life group also felt weaker positive ingroup emotions than did Yankees fans, t(48) = 2.07, p = .044, d = 0.59.

Group Identification

For the N-MB groups, importance of the ingroup was strongly correlated with positive ingroup emotions, r(57) = .60, p < .001, but considerably less so with negative outgroup emotions, r(57) = .28, p = .035. Positive ingroup emotions and negative outgroup emotions were not significantly correlated, r(57) = .22, p = .09. Using the procedure defined by Meng, Rosenthal, and Rubin (1992) for testing differences between dependent correlations, we found the difference between the correlations of importance with positive and negative emotions was significant, z = 2.31, p = .021. This suggests importance of N-MB ingroups was more strongly related to ingroup positivity than outgroup negativity. However, for MB groups, ingroup importance was equally correlated with both positive ingroup emotions, r(73) = .48, p < .001, and negative outgroup emotions, r(73) = .41, p < .001. The difference between these two correlations for MB groups was not statistically significant, z = 0.64, p = .52. The correlation between positive ingroup emotions and negative outgroup emotions was statistically significant for MB groups, r(73) = .36, p = .002. MB group importance was equally associated with both ingroup positivity and outgroup negativity, suggesting that social identification for MB groups may not be explained solely by ingroup love, but appears to involve outgroup hate as well.

We additionally conducted two multiple regression analyses on the strength of identity. In both analyses, we predicted identification from feeling thermometer ratings of both the ingroup and the outgroup, but each analysis was separated by the type of group. We expected that only the ingroup feeling thermometer, but not the outgroup feeling thermometer, would be associated with identification for the N-MB group. However, we expected both feeling thermometers would predict identification for MB groups. As expected, the proportion of variance in identification explained in each model was similar for MB (R 2 = .22) and N-MB (R 2 = .18) groups. However, the specific relationships were different across the two groups. In the model for the N-MB group, warmer feelings about the ingroup were significantly associated with stronger identification, b = 0.059, SE = 0.019, p = .002, but feelings about the outgroup was not significant, b = −0.01, SE = .01, p = .45. As predicted, both feeling thermometers were significantly related to MB ingroup identification. More negative (i.e., cold) feelings toward MB outgroups were associated with stronger identification, b = −0.037, SE = 0.010, p < .001. Additionally, more positive feelings toward MB ingroups also predicted stronger identification when controlling for outgroup feelings, b = 0.023, SE = 0.010, p = .030.Footnote 3

Summary

In sum, Study 2 replicated the results of Study 1 while also ruling out the possibility that the differences between the two types of groups could be accounted for by group status. Identification was differentially related to ingroup positivity and outgroup negativity for N-MB groups, but the strength of these correlations was not different for the MB groups. This finding therefore provides strong evidence that the nature of morality-based social identity is expressed with both positive and negative emotions. Moreover, the strength of identification with the MB group was predicted by both attitudes toward the outgroup and the ingroup, but only ingroup attitudes predicted identification for the N-MB groups. Together these analyses suggest a clear relationship between evaluations of ingroup members and identification for N-MB groups, but identification clearly involves evaluations of both groups in the case of the MB groups.

Study 3

Another alternative explanation for the effects of the prior studies is that there might be something unique about the specific groups under investigation that may have accounted for the differences found between the MB and N-MB groups. We therefore attempted to replicate the studies with different groups. Political ideology and party identifications in the US have become strongly morality-based and thus were expected to produce the same ingroup-outgroup differences found in Studies 1 and 2. Politics is essentially about the social regulation of morality (see, e.g., Janoff-Bulman, 2009). As politics has become increasingly partisan, morality and politics have become aligned, and political parties have increasingly become morality-based groups (Janoff-Bulman & Parker, 2012). Political positions are seen in terms of intentional harm perpetrated by the political outgroup. Consequently, today identification with a political party invokes the strong moral convictions associated with the political left or right. In this study we sought to replicate the first two studies with new target groups: political liberals and conservatives (i.e., morality-based) and students identified with their university (i.e., non-morality-based).

Method

Participants

Sixty-eight students at the University of Massachusetts volunteered to participate. The target groups for this study were political liberals and conservatives, or University of Massachusetts students and University of Connecticut students. Identification was measured on 9-point scales (−4 = “Strongly Disagree”; +4 = “Strongly Agree”) with two items assessing participants’ attachment to other members of their group and whether they felt they had a lot in common with other group members. Participants were only included in analyses if the average of the two items exceeded zero. Similar to Study 2, political liberals and conservatives were combined to create the MB group (n = 23), and students identified with UMass comprised the N-MB group (n = 24).Footnote 4

Measures

Participants indicated the extent to which they felt positive emotions (e.g., happiness, pride, enthusiasm) about their ingroup and negative emotions (e.g., anger, disgust, fear) about the outgroup. A condensed threat scale was created by selecting four items from the previous studies. Participants also completed feeling thermometers for the ingroup and the outgroup indicating how warm they felt about each group (0 = “Very Cold”; 100 = “Very Warm”).

Results and Discussion

The results of this study replicated the findings from Studies 1 and 2. Members of MB groups (M = 4.28, SD = 1.68) felt significantly weaker positive emotions about their ingroup than members of N-MB groups (M = 5.32, SD = 1.20), t(45) = 2.45, p = .018, d = 0.72. However, negative outgroup emotions were stronger for the MB (M = 2.80, SD = 2.00) than the N-MB group (M = 1.14, SD = 1.79), t(45) = 3.00, p = .004, d = 0.88, and threat was also significantly stronger for the MB group than the N-MB group (Ms = 0.83 vs. −2.45, respectively), t(46) = 5.64, p < .001, d = 1.65.

Building on the results from Study 2, we found additional evidence for the interdependence of positive ingroup evaluations and negative outgroup evaluations for MB (but not N-MB) groups. We conducted a moderated regression analysis on positive ingroup emotions. We used the group type (MB = 0, N-MB = 1) and the feeling thermometer rating of the outgroup (and the interaction term) as predictors. As expected, the interaction was significant, b = 0.067, SE = 0.025, p = .011 (see Fig. 2). More negative evaluations of outgroups predicted stronger positive emotions toward ingroups, but only for the MB groups, b = −0.054, SE = .020, p = .011. This pattern was not seen with N-MB outgroups; the simple slope was not significant, b = 0.013, SE = 0.015, p = .38. In any case it appeared to reflect the reverse pattern, with warmer evaluations of the outgroup predicting slightly stronger positive emotions toward the ingroup.

Fig. 2
figure 2

Positive ingroup emotions as a function of feeling thermometer ratings of the outgroup and the group type. Low (i.e., cold) and high (i.e., warm) values of the outgroup evaluation represent 1 SD below and above the mean, respectively

Full size image

The findings from this study clearly demonstrate a powerful influence of morality on emotional reactions toward ingroups and outgroups. Even with a relatively non-emotional social identity to represent the N-MB group, positive emotions toward MB ingroups were still weaker. These results seem to reflect different motivations for morality-based ingroup identification. Membership, and in turn identification with a morality-based group, may be rooted in part in the fundamental threat from the outgroup. Moreover, there seems to be a necessary interdependence between positive emotions felt toward MB ingroups and negative emotions toward MB outgroups. The nature of morality-based social identity involves both attitudes and feelings about the outgroup as well as the ingroup. In other words, how ingroups are evaluated in a morality-based context necessarily depends not only on the strength of identification but also on the extent to which the outgroup elicits negative emotions.

General Discussion

The presumed primacy of the ingroup—and its associated ingroup “love”—seems to capture the nature of the non-morality-based groups in these studies, but fails to adequately characterize the morality-based groups. The avid sports fans in Studies 1 and 2 as well as the students in Study 3 expressed stronger positive emotions for their own group than did members of the morality-based groups, but the abortion-based groups and political groups reported stronger negative emotions toward outgroups than did the sports fans or university-affiliated students. It appears that outgroups elicit stronger negative emotions in a morality-based context, whereas ingroups elicit stronger positive emotions in a non-morality-based context. The difference in outgroup-directed negativity appears to be due to the inherently threatening nature of moral (versus non-moral) outgroups.

Studies 2 and 3 also revealed an important difference between these types of groups. In Study 2, identification with the non-morality-based groups was only predicted by positive evaluations of ingroup members (but not outgroup members), but morality-based identification was predicted by evaluations of both the ingroup and the outgroup. This suggests positivity toward the ingroup and negativity toward the outgroup are dependent on one another for morality-based groups. That possibility was supported in Study 3, in which outgroup evaluations predicted positive ingroup emotions for the morality-based group, but not for the non-morality-based group. In other words, in the case of morality-based social identity, we like those who are “not them.” Together these findings imply that groups based on strong moral convictions may be different from other kinds. This type of social identity may be more outgroup-focused than is typical for group identification. No doubt reflecting the unique nature of moral convictions, moral social identities are oppositional, and ingroup positivity appears to be less a function solely of a desire to belong to the group; rather, a desire to distance oneself from the outgroup plays a crucial role as well. These findings extend our understanding of the interpersonal consequences of moral disagreements from the dyad to the group (e.g., Haidt, et al., 2003; Skitka, et al., 2005) and suggest not only that the ingroup’s reactions are threat-based but also that responses to their own group are fundamentally based in part on their intense negative responses to the outgroup.

This research suggests that outgroup “hate” plays a central role in moral social identities. Ingroup love may be the dominant form of bias for many or even most social groups, as past work has suggested (for a review, see Brown, 2000; also see Correll & Park, 2005), but to the extent that moral convictions are the salient feature of group membership and intergroup differentiation, outgroup derogation may be an equally powerful motivation for intergroup bias. What makes morality-based groups particularly unique is that the threat presented by the outgroup is an inherent feature of identification. Although some non-morality-based may feel threatened by particular outgroups, these threats must be elicited by an external event or situation. Outgroup threat is necessarily incorporated into morality-based group membership, and simply thinking about the outgroup is threatening.

These three studies represent a first step in an effort to understand the nature of morality-based groups. There may well be other reasons for the differences found between morality- and non-morality-based groups (e.g., group homogeneity, psychological essentialism). In considering potential alternatives, we were drawn to the possibility that the dichotomous, or binary, nature of the morality-based groups (i.e., pro-life versus pro-choice, Democrat v. Republican) could account for the potency of outgroup negativity. The non-morality-based groups involved rivals (Red Sox and Yankees, UMass and UConn), but each group nevertheless has multiple outgroups. Perhaps outgroup negativity is more common when there is a clear, single outgroup that is automatically elicited when thinking of the ingroup. To test this alternative, we compared morality-based groups with gender, which is also binary (male versus female). Participants (N = 66) were asked to think about a moral issue that is very important to them (or their gender), to focus on the people who agree with them on this issue (or share the same gender), and then write “what it means to be a member of this group.” Two independent coders noted whether participants mentioned an outgroup, and if mentioned, whether they discussed the outgroup negatively or not (overall agreement: outgroup mentioned, κ = .74, and outgroup negativity, κ = .67). In both the groups, slightly more than half of the participants spontaneously mentioned an outgroup (morality group: 52.5 %; gender group: 53.8 %, p = .92). However, these two groups differed considerably in how they wrote about these groups. Participants in the moral condition (61.9 %) were significantly more likely to discuss the outgroup in a negative way compared to the participants who discussed the other gender (21.4 %), χ 2(1, N = 35) = 5.55, p = .019. Thus in the case of dichotomous groups, thinking about the outgroup, although common, does not necessarily breed negativity; this negative reaction to the outgroup seems to be particularly likely in the case of morality-based groups.

The current studies also examined “pure” examples of morality-based groups. We expect that strong moral beliefs may be a part of many social identities, or perhaps more accurately, social identities can be more or less morality-based, depending on cultural history or how a group member chooses to perceive the outgroup. Thus, for example, groups based on race or religion may differ in terms of the degree to which intergroup differences are regarded as morality-based. The studies’ findings suggest that to the extent that differences are perceived as based in morality, outgroup negativity and derogation are likely to increase, and in turn ingroup positivity is likely to be strengthened. We are currently exploring these possibilities.

For the morality- and non-morality-based groups in this research very different patterns of ingroup positivity and outgroup negativity arose. Important directions for future research lie in further exploring and specifying the role of morality in social identity and intergroup conflicts. Other differences may exist between morality- and non-morality-based groups that would have impacts on intergroup relations. The current research found different patterns of threat and group-based emotions for morality-based social identities and suggests that an understanding of morality and moral convictions may help inform the study of intergroup relations.