Agentic and Communal InteractionGoals in Conflictual IntergroupRelations

Members of conflicting groups experience threats to different identity dimensions, resulting in the need to restore the aspect of identity that was threatened. Do these needs translate into specific goals in social interactions? In the present research, we examined the hypotheses that (1) experiencing one’s ingroup as illegitimately disadvantaged or victimized arouses agentic goals (to act and appear assertive and confident) when interacting with the advantaged or victimizing group, while (2) experiencing one’s ingroup as illegitimately advantaged or perpetrating transgressions arouses communal goals (to act and appear warm and trustworthy) when interacting with the disadvantaged or victimized group. Study 1 (N = 391) generally supported both hypotheses across diverse intergroup contexts involving gender, national/ethnic, and consumer identities. Study 2 (N = 122) replicated this pattern in a context of occupational identities. Study 2 further showed that the effect of ingroup role on agentic and communal intergroup goals was not moderated by participants’ general dispositional preferences for agentic and communal goals in interpersonal interactions, thus demonstrating how ingroup role exerts a distinct and robust influence on goals for interactions with other groups. Theoretical and practical implications are discussed.


Background
How people want to interact (e.g.acting friendly or acting dominant) with members of other groups may not only be influenced by their personality (i.e., stable preferences for certain kinds of behavior), but also by the social role of their ingroup.Especially when groups are in conflict, their members should be influenced by their group's role of being either a victim or disadvantaged, or being a perpetrator or illegitimately advantaged when interacting with the respective other group.

Why was this study done?
With our work, we build on previous research which provided preliminary support for the assumption that group members who feel that their group is treated unfairly or even victimized by another group will try to regain power and control by acting dominantly, whereas group members who feel that their group has harmed another group or enjoys illegitimate advantage will act in friendly ways in order to be accepted by the other group despite their immoral actions or illegitimate advantage.Since previous research tested these assumptions only indirectly or only for very specific situations, we aimed to show that very different groups in very different contexts may pursue similar goals as soon as they feel that their group is either victimized/disadvantaged or perpetrating transgressions/advantaged. of the involved parties during a conflict: "the view of one's self as weak and vulnerable, and the view of one's self as violent and unjust" (p.593).This twofold conceptualization of feeling weak and vulnerable on the one hand or violent and unjust on the other hand aligns with the two fundamental dimensions of social cognition (the "Big Two") -namely, (deficient) agency, which refers to competence, assertiveness, and decisiveness, and communion, which refers to benevolence, trustworthiness, and morality (Abele & Wojciszke, 2014;Judd, James-Hawkins, Yzerbyt, & Kashima, 2005).
The parties involved in a conflict typically differ in the extent to which they view themselves as weak and vulnerable, that is, experience a threat to their agentic identity, or as violent and unjust, that is, experience a threat to their communal identity.Specifically, according to the needs-based model (Nadler & Shnabel, 2008, 2015), a group's social role within a conflict determines the specific type of identity threat experienced by its members.Originally developed for contexts characterized by direct violence (i.e., where conflicting groups concretely aggress against each other ;Shnabel, Nadler, Ullrich, Dovidio, & Carmi, 2009), the model posits that victims and perpetrators experience different threats to their ingroup's positive identities following transgressions: Whereas members of victimized groups experience threat to their agentic identity -namely, sense of power, control, and self-determination, members of perpetrator groups experience threat to their communal identity -namely, sense of being morally adequate actors, who are socially accepted by their community (as social exclusion is the sanction imposed upon those who violate moral standards; Tavuchis, 1991).
Similar threats are assumed for members of disadvantaged and advantaged groups in context of structural inequality (i.e., contexts that are characterized by "systematic inequalities in the distribution of economic and political resources", Christie, 1997, p. 323).Consistent with this idea, the stereotype content model (Fiske, Cuddy, Glick, & Xu, 2002) posits that disadvantaged groups are often perceived as warm but incompetent (representing a threat to their agentic identity), whereas advantaged groups are often perceived as competent but cold and immoral (representing a threat to their communal identity).Importantly, members of advantaged and disadvantaged social groups seem to experience these divergent identity threats especially when they perceive the disparity between groups as illegitimate (Shnabel & Ullrich, 2013;Siem, von Oettingen, Mummendey, & Nadler, 2013).If status differences are perceived as normative and justified, then intergroup relations can be described as non-conflictual and group-based disparities should be experienced as less threatening to the social identities of advantaged and disadvantaged group members (e.g., Caddick, 1982;Tajfel, 1978).Perceiving that one's group is in an advantaged or disadvantaged position for illegitimate reasons, however, should make group members vulnerable to social identity threats (e.g., Tajfel & Turner, 1979) with members of advantaged groups likely experiencing a threat to their communal identity, and members of disadvantaged groups likely experiencing a threat to their agentic identity.Supporting this reasoning is the finding that in interracial (but not intra-racial) interactions black and Latino-Americans (lower status groups in the U.S.) were motivated to be respected, whereas white Americans (a higher status group) were motivated to be liked (Bergsieker, Shelton, & Richeson, 2010).
Obviously, contexts of direct violence and structural inequality differ in many respects, such as the nature and severity of transgressions.Being wounded in a military operation is not the same as being viewed as incompetent and discriminated against in an employment setting.Nevertheless, the distinction between the two types of contexts is not always clear-cut (e.g., do police shootings of unarmed black Americans represent direct violence or structural inequality?).Moreover, according to the Big Two theorizing (Abele & Wojciszke, 2014), in both types of contexts the threats posed to the identities of the involved groups fall under the overarching categories of agency (for victim and disadvantaged groups) and communion (for perpetrator and advantaged groups) (see SimanTov-Nachlieli, Shnabel, & Nadler, 2013).
According to social identity theory, feeling threatened regarding one's positive social identity results in the motivation to overcome this threat and restore a positive self-concept (Tajfel & Turner, 1979).Building on this idea, the needsbased model (Nadler & Shnabel, 2008, 2015) highlights that for a positive identity, both the agentic and the communal identity dimensions need to be unscathed.One way of restoring a specific threatened identity-dimension is to highlight characteristics of one's ingroup that are associated with this dimension in the interaction with the respective outgroup.For example, being motivated to restore the ingroup's agentic identity, one might want to act and appear assertive, decisive, and confident, whereas being motivated to restore the ingroup's communal identity, one might want to show concern for the other parties' welfare or willingness to compromise.This reasoning accords with self-determination theory (Deci & Ryan, 2000) which posits that people generally strive to satisfy their basic psychological needs in their actions and are motivated to behave in ways that they perceive as reflecting who they want to be.In the present research we focused on these agentic and communal "interaction goals" and how they are affected by the ingroup's perceived social role of being either illegitimately disadvantaged or victimized, or illegitimately advantaged or perpetrating, respectively.
There is some indirect evidence suggesting that members of victimized and perpetrating groups pursue different interaction goals.To illustrate, in a study assigning participants' ingroups to the social roles of either victim or perpetrator (e.g., by reminding Jewish and German participants of the Holocaust, see Shnabel et al., 2009, Study 2), participants read messages allegedly sent by representatives of the respective other party.Those messages reaffirmed the participants' ingroup's agency or communion: Whereas the agency-reaffirming message mentioned participants' ingroup's "right to be strong and proud in their country and have the power to determine their own fate", the communion-reaffirming message highlighted moral-social acceptance for participants' ingroup and understanding that "it is not easy for [participants' ingroup] to live with the past".From perpetrators' more positive reactions (i.e., higher willingness to reconcile with the outgroup) to communion-reaffirming compared to agency-reaffirming messages it was concluded that they experienced a heightened need to restore their communal identity, whereas from victims' more positive reactions to agency-reaffirming (compared to communion-reaffirming) messages it was concluded that they experienced a heightened need to restore their agentic identity.The same approach was successfully used in contexts of structural inequality, referring to (Study 1) or manipulating (Study 2) the status of student participants' universities by pointing to unequal chances for admission to graduate programs (Shnabel, Ullrich, Nadler, Dovidio, & Aydin, 2013).
An advantage of this technique is that it points to a useful intervention in conflictual intergroup relations: Sending identity-restoring messages can ameliorate identity threats, which in turn can improve intergroup relations.A limitation of this approach is that the assumed needs have only been inferred from participants' responses to agencyor communion-reaffirming messages.Making such an inference is problematic, given the distinction between processes of identity definition -cognitively defining oneself as a symbolic object with particular characteristics and descriptive labels, and identity enactment -behaviorally acting out aspects of one's identity (Vignoles, 2011).
For example, perceiving oneself as a powerless victim may lead to a heightened need for agency (as predicted by the needs-based model), but at the same time to non-agentic (submissive, unassertive) behavior that is consistent with the labeling of oneself as a victim (see for example, inertia effects following the experience of humiliation; Ginges & Atran, 2008).Hence, a direct empirical examination of group members' interaction goals is necessary.
Preliminary evidence for an effect of social role on people's communal and agentic interaction goals comes from two experiments conducted by Siem et al. (2013) in contexts of structural inequality.Both experiments manipulated status differences and their legitimacy.The first experiment used a minimal group paradigm, while the second experiment referred to status differences between real world groups, namely clinical psychologists and social workers (a lower status group in this occupational context) or psychiatrists (a higher status group).In the study that used the minimal group paradigm, illegitimacy was manipulated by giving one group an unfair advantage (i.e., permission to use a calculator in a math test) over the other (vs.equal conditions); in the study that used real world groups, illegitimacy was manipulated by describing the status differences as unfair (e.g. by referring to the similar tasks conducted by the different professional groups) or fair (e.g. by referring to the different training of the different professional groups).As expected, only in the illegitimacy conditions did disadvantaged group members express a higher need for agency and a lower need for communion compared to advantaged group members.While these studies provide tentative support for our assumptions, they are not without limitations.Specifically, they relied on ad-hoc created, study-specific measures of needs, and focused on only two different intergroup contexts (one of which was a somewhat artificial minimal group context).For contexts of direct violence, there is no direct evidence for the effect of social role on interaction goals at all.In summary, our work builds on previous research which showed only indirectly that members of victimized and perpetrator groups experience different goals (to be agentic or communal, respectively) following a conflict depending on their group's social role (Shnabel et al., 2009) and on research which provided preliminary support for the notion that members of illegitimately (dis-)advantaged groups experience very similar goals, again, depending on the social role of their group (Siem et al., 2013).With the present research, we aimed to test the robustness of the supposed effect by examining two questions: First, does the effect of social role on agentic and communal goals occur in different contexts (which, however, all meet the prerequisites for the applicability of the needs-based model hypotheses, that is illegitimate harm or inequity), and second, does the effect of social role on agentic and communal goals occur regardless of individuals' habitual interpersonal interaction goals (which may be opposed to the interaction goals individuals are predicted to pursue as members of a specific group).

The Present Research
We conducted two studies using different contexts to submit our main hypotheses, derived from the needs-based model (Nadler & Shnabel, 2015), to two different critical tests.Reflecting the importance of systematic conceptual replications for external validity and construct validity (Shadish, Cook, & Campbell, 2002), the first test addressed the generality of the model's prediction that -in intergroup interactions -members of illegitimately disadvantaged or victim groups will have stronger agentic but weaker communal goals than members of the corresponding illegitimately advantaged or perpetrator groups.Previous research using ad-hoc measures has provided initial support for this prediction in a context of structural inequality (Siem et al., 2013).Study 1 extended this research by systematically testing the influence of ingroup role on agentic and communal interaction goals across five highly diverse contexts of intergroup transgressions or illegitimate inequality using a rigorous, previously validated measure of interaction goals (i.e., the Circumplex Scales of Intergroup Goals; Locke, 2014).
Study 2 aimed to replicate and extend Study 1. Specifically, Study 2 tested whether and how the effect of the ingroup's role (i.e., illegitimately disadvantaged or advantaged) on group members' goals for intergroup interactions is influenced by their general preferences for pursuing agentic and communal goals in their interpersonal interactions.Social categorization theory predicts that when membership in a particular social group becomes salient, a process of depersonalization occurs such that people adopt that group's norms and values (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987).Thus, in intergroup contexts that highlight group membership, one's intergroup goals should be determined by the ingroup's role in that situation.However, the question remains as to whether, beyond this main effect of ingroup role suggested by self-categorization theory and the needs-based model, one's personal disposition to pursue agentic or communal goals moderates the effect of ingroup role.To examine this question, which has never been empirically tested before, we assessed participants' dispositional tendencies to seek agency and communion in interpersonal interactions before manipulating their ingroup's social role.Furthermore, as a partial replication of Siem et al.'s (2013) work, Study 2 experimentally manipulated the (il)legitimacy of status differences.We expected to replicate the described effect of social role on agentic and communal goals only if status differences are perceived as illegitimate.
Together, the present studies aimed to (a) provide a systematic and comprehensive account of how intergroup interaction goals depend on which roles groups occupy within contexts of illegitimate intergroup inequality and transgressions, and (b) explore whether and how goal pursuit at the interpersonal level affects goal pursuit at the group level.

Study 1
Using participants' actual membership in various "real world" groups, Study 1 tested the prediction that participants reminded of their membership in an illegitimately disadvantaged or victim group would prefer their group pursue more agentic but less communal interaction goals than participants reminded of their membership in the corresponding illegitimately advantaged or perpetrator group across five diverse intergroup contexts involving gender, national/ethnic, and consumer identities.

Method Participants and Design
Psychology students (N = 404) at a large German public distance teaching university voluntarily participated in an online survey in exchange for course credit.Thirteen participants were excluded from analyses because they either did not complete the survey or indicated that they did not respond seriously; thus, the final sample included 391 participants (304 female, 87 male, M age = 32.22,SD = 9.56).While 282 participants were of German descent, 109 participants indicated that either they or their parents or grandparents were immigrants.
Participants' nationality/ethnicity and gender were used to assign them to the different cells of the 2 (Role [illegitimately disadvantaged/victim, illegitimately advantaged/perpetrator]) × 5 (Context [1,2,3,4,5]) between-subjects design i .Contexts referred to discrimination of immigrants in Germany (Context 1), right-wing violence (Context 2), an international transgression (Context 3), exploitative consumption (Context 4), and gender discrimination (Context 5) (for detailed descriptions of the context variable and its operationalization, see below; for a verbatim description, see Table A1a).In the discrimination of immigrants context, 51 migrant participants were assigned to the illegitimately disadvantaged condition (41 female, 10 male) and 36 non-migrant participants to the illegitimately advantaged condition (28 female, 8 male).In the right-wing violence context, 58 migrant participants were assigned to the victim condition (49 female, 9 male) and 40 non-migrant participants to the perpetrator condition (31 female, 9 male).In the remaining conditions, which only included non-migrant participants, there were 28 participants in the victim condition (21 female, 7 male) and 32 participants in the perpetrator condition (24 female, 8 male) in the international transgression context, 30 participants in the victim condition (22 female, 8 male) and 41 participants in the perpetrator condition (32 female, 9 male) in the exploitative consumption context, and 30 female participants in the illegitimately disadvantaged condition and 11 male participants in the illegitimately advantaged condition in the gender discrimination context.
Note that while Contexts 1, 2 and 5 used participants' membership in natural groups (i.e., immigrants vs. nonmigrants; women vs. men) and should therefore be considered as quasi-experimental, in Contexts 3 and 4 all participants were randomly assigned to a condition that presented their ingroup as either illegitimately disadvantaged/victimized or illegitimately advantaged/perpetrating transgressions. ii

Procedure and Measures
After providing demographic information, participants read short one-sentence statements which reminded them of their ingroup's role in the various contexts.The contexts referenced several controversies in the German media involving conflictual intergroup relations characterized by illegitimate group inequality or transgressions.Following the manipulation, participants completed the following measures using 5-point scales (1 = not true at all/not important to 5 = completely true/very important).At the end of the survey, participants were fully debriefed.

Role and context manipulation -
The intergroup conflicts used in all five contexts were heavily covered by the German media at the time we collected the data (summer 2014).Even though the five contexts were very diverse, the manipulation texts were formally identical.Each consisted of two sentences of the following form: "As you know [clarification of relationship between the participant's ingroup and the outgroup; e.g., "immigrants are disadvantaged compared to Germans"].How should we [participant's ingroup; e.g., "Germans"] behave towards [respective outgroup; e.g., "immigrants"] in your opinion given this fact?"Because the needs-based models' hypotheses can be transferred from contexts of direct violence to contexts of structural inequality only when the groups' inequality is perceived as illegitimate, contexts referring to group inequality involved inequalities we expected most participants would perceive as illegitimate.Table A1a presents the verbatim manipulations in all conditions.Below, for the benefit of readers who are unfamiliar with the five contexts that we used, we describe each of them in more detail.Note, however, that the participants read only the short sentence described above and not the detailed descriptions below.
The first context referenced the relationship between immigrants and the non-migrant majority in Germany.The manipulation text reminded participants of the general structural disadvantage of immigrants compared to nonmigrants, thus assigning non-migrant participants to the advantaged role and immigrant participants to the disadvantaged role by using the identical manipulation text in both role conditions.At the time we collected the data, the discrimination of immigrants -especially in the job market -was heavily discussed in the German media (e.g., Lüpke-Narberhaus [2014, March 26: "Keiner will einen Ali im Team haben"/"Nobody wants to work with Ali"]; "Joe Biden nennt Deutschland ausländerfeindlich"/"Joe Biden calls Germany xenophobic" [2014, June 11]).
The second context referred to the concrete transgressions of the far-right German terrorist group "Nationalsozialistischer Untergrund" (NSU; national-socialist underground) that murdered nine immigrants from Turkey and Greece between 2000 and 2006.In the course of the trial against NSU, massive investigative blunders by German authorities were revealed and intelligence agencies were criticized for underestimating the threat of right-wing extremism and for failing to convey its findings more publicly.Importantly, the manipulation text in that condition referenced the investigative shortcomings rather than the NSU murders itself, because we assumed that most German participants would identify with German authorities (and feel guilty for their failures) but not with the terrorist group.As in the first context, non-migrant participants were assigned to the perpetrator role and immigrant participants to the victim role by using the identical manipulation text in both role conditions.
The third context reminded (non-migrant) German participants of international transgressions in which Germany assumed either a victim or a perpetrator role.The victim condition involved the United States National Security Agency (NSA) espionage scandal during which the NSA bugged Angela Merkel's mobile phone.The perpetrator condition involved Germany's refusal to pay reparations for 200 Greek civilians who were killed by the Nazis during WWII.Similar to the second context, this condition referenced German authorities' refusal to pay the reparations and not the killings itself.We acknowledge that the situations in the two conditions of Context 3 differ in several ways beyond the social role of the ingroup.Nevertheless, they are comparable in that both cases referred to a concrete transgression involving another nation (i.e., USA or Greece).
The fourth context asked German participants to think about their identity as consumers.In the victim condition, they were reminded of a recent media revelation that Google illegally used information involuntarily provided by its consumers.Thus, this condition addressed a potential threat to participants' control (over their privacy) and exposure to unjust treatment, constituting a victim position.In the perpetrator condition, participants were reminded of the consequences, in terms of inhumane working conditions, caused by their consumption of cheap textile products from low-wage countries.As in Context 3, victim and perpetrator conditions differ in several ways, but are nevertheless comparable in that both cases referred to consumer identities.
Finally, the fifth context was identical for both advantaged (men) and disadvantaged (women) roles and concerned women unfairly earning less than men showing the same performance.
To make sure that the participants understood the referenced contexts and the role their ingroup played, we asked them to indicate their knowledge about it using a 5-point scale ranging from 1 = no knowledge at all to 5 = very good knowledge.In all contexts, participants' knowledge varied around the scale's midpoint, reflecting "average knowledge".The only exception was Context 3 (Germans' refusal to pay reparations for perpetration in Greece during WWII), in which the average score was below the midpoint (M = 1.91,SD = 0.82), reflecting only "little knowledge." Manipulation check -Participants indicated their agreement with two items assessing perceptions of the illegitimately advantaged/perpetrator role ("In this context, I see my group as having the perpetrator role," "My group has acted immorally in this context"; r(354) = .66,p < .001)and two items measuring perceptions of the illegitimately disadvantaged/victim role ("In this context, I see my group as having the victim role," "My group has been treated unjustly in this context"; r(346) = .70,p < .001).In order to create one manipulation check index, we subtracted the averaged illegitimately disadvantaged/victim score from the averaged illegitimately advantaged/perpetrator score so that negative values reflect greater salience of the illegitimately disadvantaged/victim role (i.e., feeling part of a victimized or illegitimately disadvantaged group) and positive values reflect greater salience of the illegitimately advantaged/perpetrator role (i.e., feeling part of a perpetrator or illegitimately advantaged group).
Because perceiving the inequality between the groups as illegitimate is a precondition for the applicability of hypotheses derived from the needs-based model (Siem et al., 2013), we chose contexts that we expected most people would perceive as involving illegitimate harms or inequities.To check whether the precondition of illegiti-macy perceptions was met, we assessed perceived legitimacy of the described contexts with a single item using a 5-point scale (1 = not legitimate to 5 = absolutely legitimate; see Table A1b for the items, which were adapted to each respective study context).
Interaction goals -To assess agentic and communal interaction goals, we used the Circumplex Scales of Intergroup Goals (CSIG; Locke, 2014), a 32-item measure that assesses a diversity of goals reflecting all possible mixtures of agentic and communal tendencies and that has been successfully applied to a variety of intergroup contexts (Aydin, Ullrich, Siem, Locke, & Shnabel, 2018;Locke, 2014;Locke & Heller, 2017;SimanTov-Nachlieli, Shnabel, Aydin, & Ullrich, 2018).The following text preceded the CSIG items, and varied as a function of experimental condition: "When we [ingroup] interact with [outgroup], it is important to me that… e.g."…we show concern for their welfare", "…we understand their point of view" (example communal items) or "…we are assertive", "…we appear confident" (example agentic items).Participants rated the importance of each goal on 5-point scales ranging from 1 = not important to 5 = very important.
The items were translated into German by two independent translators; the final version was back-translated by a native speaker, reaching a very high correspondence with the original version, verified by its author.Agentic and communal dimension scores theoretically range from -4 to +4; in Study 1, participants' agency scores ranged from -2.14 to 2.65 (M = 0.25, SD = 0.78) and their communion scores ranged from -0.56 to 4.00 (M = 1.76,SD = 0.83).The reliabilities of the agentic and communal dimensions were .83and .88,respectively, and the CSIG's fit to a circumplex model was significant (that is, items representing similar goals were more strongly positively correlated than items representing different goals.For detailed information on how we tested the circumplex structure and calculated agentic and communal dimension scores, see https://osf.io/zatfb/).

Results
The data (https://osf.io/zexg3/)and the syntax (https://osf.io/pqncx/)we used in Study 1 are available at the Open Science Framework.See Table A2a for descriptive statistics and bivariate correlations between the independent and main dependent variables used in Study 1.As explained above, our aim was to probe the generalizability of the social role effect on interaction goals across different intergroup contexts, which would be statistically reflected in a significant main effect of social role and either a non-significant or an ordinal interaction between role and context.In an ordinal interaction, the magnitude of the social role effect would vary across the different context levels, but -in contrast to a disordinal interaction -the direction of the social role effect would be the same at all context conditions (see Aiken & West, 1991).In order to test this interaction, the data from all five contexts were combined in one analysis.A further advantage of this joint analysis of Contexts 1-5 is the higher statistical power to detect the expected effect of role on agentic and communal interaction goals.We used Bonferroni correction to compensate for the increased probability of falsely rejecting the null hypothesis when carrying out multiple post hoc comparisons.

Manipulation Check
To check the effectiveness of the role manipulation, we subjected our manipulation check index to a 2 (Role disadvantaged/victim conditions were negative, indicating a greater salience of the illegitimately disadvantaged/victim role, whereas all means in the illegitimately advantaged/perpetrator conditions were positive, indicating a greater salience of the illegitimately advantaged/perpetrator role.Differences between the groups were all significant, except in Context 1 (see Table 1), where our manipulation failed to induce different role perceptions in illegitimately disadvantaged and advantaged participants (immigrants and non-migrants, respectively).Note.The disadvantaged conditions subsume illegitimately disadvantaged and victimized roles, and the advantaged conditions subsume illegitimately advantaged and perpetrator roles.Due to Bonferroni correction, only simple effects test with a p-value less than .01are considered significant.Effect sizes are based on the root of the mean squared error from the ANOVA, which puts all simple main effects on a common metric.
We also checked whether the precondition of illegitimacy perceptions was met.With scores ranging from M = 1.23,SD = 0.43 to M = 2.88, SD = 0.97 on a 5-point scale, participants in all conditions perceived the described contexts as rather illegitimate.In eight of ten conditions, perceived legitimacy was significantly below the scale midpoint, the exceptions being immigrant participants assigned to the victim condition in Context 2 (failure to investigate right-wing violence context) and non-migrant German participants assigned to the perpetrator condition in Context 3 (transgression against Greek civilians context).Table 2

Agentic Interaction Goals
To examine the effect of role and context on agentic interaction goals, a 2 × 5 ANOVA was conducted.The results revealed significant main effects of role, F(1, 347) = 142.38,p < .001,η 2 = .29,and context, F(4, 347) = 7.00, p < .001,η 2 = .08.Furthermore, the two-way interaction was significant, F(4, 347) = 8.77, p < .001,η 2 = .09,indicating that the effect of role varied across contexts.Figure 1 illustrates the nature of this ordinal interaction (in which the effect of role, albeit in different size, was in the same direction in all five contexts).All post hoc comparisons were in the expected direction (see Table 3 for means and standard deviations) and all were significant except for Context 1 (discrimination of immigrants).That is, participants reminded of the illegitimately disadvantaged/victim role of their ingroup expressed stronger agentic interaction goals than participants reminded of the illegitimately advantaged/perpetrator role of their ingroup.Removing the participants who scored above the midpoint of the perceived illegitimacy scale (n = 39, and two participants with missing values) did not change the pattern of results.Note.The disadvantaged conditions subsume illegitimately disadvantaged and victimized roles, and the advantaged conditions subsume illegitimately advantaged and perpetrator roles.Study 1: df = 347.Due to Bonferroni correction, only simple effects test with a p-value less than .01are considered significant.Effect sizes are based on the root of the mean squared error from the ANOVA, which puts all simple main effects on a common metric.Study 2: df = 120.Effect sizes are determined by calculating the mean difference between un-residualized agency and communion intergroup scores of advantaged and disadvantaged participants and dividing the result by the pooled standard deviation.a Removing the participants who scored above the midpoint of the perceived illegitimacy scale led to a significant contrast between victimized and perpetrator participants with regard to communal goals in Context 2 (victim condition: M = 1.90, SD = 0.94, perpetrator condition: M = 2.36, SD = 0.70; t(306) = -2.81,p = .005).
In the other three contexts (i.e., discrimination of immigrants [Context 1], failure to investigate right-wing violence [Context 2], and gender discrimination [Context 5]), even though the effects were in the expected direction, illegitimately advantaged/perpetrator and illegitimately disadvantaged/victim participants did not differ significantly with regard to communal intergroup goals.Figure 1 depicts the effect of social role on communal intergroup goals in all contexts.Removing the participants who scored above the midpoint of the perceived illegitimacy scale slightly changed the pattern of results in favor of our hypothesis, in the sense that the difference in communal goals between immigrant participants (i.e., the victimized group) and non-migrants participants (i.e., the perpetrator group) was also significant in Context 2 (i.e., failure to investigate right-wing violence, see Table 3).

Discussion
The results of Study 1 generally supported our theorizing.As hypothesized, participants generally reported stronger agentic interaction goals when their ingroup occupied an illegitimately disadvantaged/victim role, and they tended to show stronger communal interaction goals when their ingroup occupied an illegitimately advantaged/perpetrator role.Social role showed predominantly large effects on agentic goals, the effect on communal goals varied de-pending on the context: While all comparisons were in the expected direction, only two out of four comparisons were significant after Bonferroni correction, yielding large effects.
In four of five contexts, we effectively induced a sense of illegitimate disadvantage or victimization, or illegitimate advantage or perpetration.The stronger the role manipulation's impact on participants' experience of their ingroup as illegitimately advantaged or perpetrator versus disadvantaged or victim (as reflected in the effect sizes of the manipulation check), the stronger its impact tended to be on participants' agentic and communal motives.In Context 1, where our role manipulation failed (i.e., immigrant participants did not perceive their ingroup as "treated unfairly" or "victimized", and non-migrant participants did not perceive their ingroup as "immoral" or having a perpetrator role, compared with the respective outgroup), we likewise failed to induce changes in agentic and communal goals -a result that is congruent with our general reasoning.
Our findings are valuable theoretically because they show that across very diverse contexts, strategies to strive for a positive self-concept (by either highlighting agentic or communal aspects of the ingroup's identity) critically differ between illegitimately disadvantaged or victimized and illegitimately advantaged or perpetrating groups.
This finding can be considered as complementing social identity theory (Tajfel & Turner, 1979) by showing that illegitimately disadvantaged or victim and advantaged or perpetrator groups all suffer inferiority in social comparisons, albeit on different identity dimensions, namely agency and communion, respectively.Research on social identity theory has previously considered the case of perceived inferiority of high status groups on status-irrelevant dimensions (e.g., Terry, 2001), but the possibility that such an identity threat would translate into communal interaction goals given perceived illegitimacy has received little attention.
The practical value of these results is that knowing how those groups typically behave towards each other can help guide structured encounters between conflicting groups, such as "dialogue group" interventions based on the contact hypothesis (Allport, 1954).Typically, such interventions focus on increasing intergroup mutual acceptance (e.g., by encouraging personalized contact aimed at friendship formation; Dixon, Durrheim, & Tredoux, 2005).Our findings suggest that this focus may satisfy the goals of members of perpetrator or illegitimacy advantaged groups, but leave the goals of members of victim or disadvantaged groups unaddressed.This may explain why members of victim or illegitimately disadvantaged groups tend to be less satisfied from intergroup contact interventions (Tropp & Pettigrew, 2005).As such, our findings offer valuable insights to practitioners planning such encounters, suggesting that they should include components that allow the expression of agentic, not just communal goals.Furthermore, our findings raise the question of how illegitimately disadvantaged or victimized groups will respond to communal behavior shown by the illegitimately advantaged or perpetrator group, and how members of the latter group respond to agentic behavior shown by illegitimately disadvantaged or victimized groups -an important direction for future research, since relations between groups who share a common space are likely to be shaped more by the "natural" behavior of outgroup members than by the exchange of thoughtful messages by outgroup representatives.
Unexpectedly, non-migrant participants did not show stronger communal tendencies compared to immigrants in the contexts of structural inequality (discrimination of immigrants, Context 1) and a direct transgression (failure to investigate right-wing violence, Context 2).However, these non-significant differences do not seem to be due to un-communal non-migrants in these contexts, but rather highly communal immigrants.In those two contexts, the illegitimately disadvantaged/victim group (i.e., immigrants) yielded a different pattern of results compared to illegitimately disadvantaged/victim participants in Contexts 3-5 (i.e., international transgression, exploitative con-sumer relations, gender discrimination).Specifically, whereas illegitimately disadvantaged/victim participants in Contexts 3-5 showed a relatively balanced ratio of agentic and communal goals, illegitimately disadvantaged/victim participants in Contexts 1 and 2 (i.e., immigrant participants) clearly prioritized communion over agency (notwithstanding the increased agentic intergroup goals in the context of right-wing violence).Apparently, it was paramount for those participants to highlight their communal identity dimension when interacting with members of the nonmigrant German majority.A possible explanation is that "the generic immigrant" is perceived as both low in agency and low in communion (Lee & Fiske, 2006).Participants with a migration background may be intuitively aware of this stereotypical view of them and thus feel threat over both core dimensions of their identity.
A potential limitation is the small sample size of some of the sub-samples.Post-hoc power analyses revealed that the statistical power to detect at least one out of ten effects assumed to be large (d = .80)was 99%.However, the power to detect all ten effects (i.e., the effect of role on both agentic and communal intergroup goals in five studies) was only 7%.Thus, it is possible that we have missed some of the effects even if they were large.In the light of these considerations of statistical power, we refrain from putting too much emphasis on the absence of any particular effect.Instead we acknowledge that our non-significant results are inconclusive.For example, given that especially the sub-sample of male participants in Context 5 was very small (n = 11), replicating the gender comparison with a reasonably large sample will be necessary to conclude whether or not agentic and communal goals differ between men and women as hypothesized.
Another limitation may be seen in the incommensurability of Contexts 3 (international transgression) and 4 (exploitative consumption) with regard to the different social roles.As a consequence, there may have been other variables (besides the social role of one's ingroup) that influenced the results (see the General Discussion for a more detailed analysis).Indeed, the main effect of context on both agentic and communal interaction goals and the significant interactions between role and context suggest that participants responded differently to the various context conditions.Nevertheless, and most important for our purpose, the effect of social role on agentic and communal goals was in the predicted direction across all studies.Similarly, although one might be concerned that we subsumed social identities that were characterized by either illegitimate (dis-)advantage (i.e., structural inequality) or perpetrator-and victimhood (i.e., direct violence) across highly diverse contexts, the consistency of the overall pattern of results supports our reasoning and offers compelling evidence of the generalizability of the hypothesized role effect on intergroup goals.

Study 2
Study 2 focused on the effects of one's ingroup's role on interaction goals in a context of structural inequality, and pursued two main aims.The first aim was to test the combined influence of both individual (i.e., "self") and group (i.e., "social") identity on agentic and communal intergroup interaction goals.Thus, in Study 2 we measured preexisting preferences to pursue agentic or communal interpersonal goals before manipulating participants' ingroup's role.This allowed us to examine how goals dispositionally pursued on the individual level (i.e., "In general it is important to me to act…e.g.friendly") relate to interaction goals pursued between groups in a particular context (i.e., "When [ingroup] interacts with [outgroup], it is important to me that we act…e.g.friendly") and whether dispositional interpersonal goals moderate the impact of group role on intergroup interaction goals (e.g., is the effect of experiencing one's ingroup as illegitimately disadvantaged on agentic interaction goals stronger for members who generally prefer to be agentic?).Note.The social role manipulation was effect-coded (disadvantaged = -1, advantaged = 1), the legitimacy manipulation was effect-coded (illegitimate = -1, legitimate = 1), interpersonal goals were z-standardized.*p < .05. **p < .01.
Figure 1 depicts the agentic and communal intergroup interaction goals of participants assigned to the disadvantaged and advantaged groups in Study 2.

Discussion
Consistent with social categorization theory (Turner et al., 1987), Study 2 demonstrated that in a context of illegitimate intergroup disparity, participants' agentic interaction goals were determined by their ingroup's situational Aydin, Ullrich, Siem et al. 161 role but not by their dispositional goals when interacting with other individuals.In line with Study 1 and the needsbased model (Nadler & Shnabel, 2015), perceiving the ingroup as illegitimately disadvantaged (vs.advantaged) evoked agentic interaction goals.Unexpectedly, the effect of social role on communal interaction goals was not significant, such that perceiving the ingroup as illegitimately advantaged (vs.disadvantaged) evoked only a tendency toward stronger communal interaction goals, perhaps because people are already highly motivated to behave communally in contexts of collegial relations (see the high scores for communal goals in Study 2 compared to the contexts in Study 1 in Figure 1).Still, this result is consistent with Study 1, in which social role had more influence on group members' agentic goals than their communal goals -an issue that we further consider in the General Discussion.
Interestingly, participants' dispositional preferences had a stronger effect on communal than on agentic intergroup interaction goals.This finding is consistent with Locke (2014) who reported higher correlations between interpersonal and intergroup goals for the communal dimension, r(122) = .49,p < .001,than for the agentic dimension, r(122) = .12,p = .174.Explaining this difference remains a topic for future research.
Unexpectedly, the legitimacy manipulation did not moderate the results of social role as it did in Siem et al. (2013).
As reported above, even though the analysis revealed a significant main effect of this manipulation on the legitimacy manipulation check, perceived legitimacy scores in all conditions were on the illegitimate side of the scale (i.e., significantly below the scale midpoint).Whereas the sample used by Siem et al. (2013) was drawn from a university which offers only one of the relevant study majors (i.e., psychology), participants of the present study were students of a university that offers all three study majors (i.e., psychology, medicine, and social work), and their exposure to students from these other majors may have made it more difficult to convince them that a legitimate status disparity could exist between these groups.Whatever the reason, based on this sample, we cannot address the question as to whether perceived legitimacy of status differences (low vs. high) acts as a moderator of the social role effect on interaction goals as shown by Siem et al. (2013).
Interestingly, we did observe an unexpected interaction between the legitimacy manipulation and the dispositional preference for communal interpersonal interactions, such that participants with high (vs.low) preference for interpersonal communion showed an increase in communal intergroup goals when assigned to the illegitimate condition (irrespective of whether the illegitimate disparity between their ingroup and the outgroup advantaged or disadvantaged them).One possible explanation is that illegitimate status disparities is the type of potentially conflictual situation that heightens communal persons' inclination to strive for harmonious relationships and heightens uncommunal persons' inclination to become guarded and self-protective.

General Discussion
The present research systematically examined how group members' pursuit of interaction goals depends on their ingroup's role as illegitimately disadvantaged or victimized versus illegitimately advantaged or perpetrating transgressions.Drawing on the needs-based model's theorizing (Shnabel & Nadler, 2008) and initial findings (Siem et al., 2013), two studies demonstrated that interaction goals vary as a function of the ingroup's social role; specifically, members of illegitimately disadvantaged and victim groups showed stronger agentic goals and weaker communal goals compared to members of illegitimately advantaged and perpetrator groups.Thus, although sometimes disadvantage and victimization can evoke helplessness (e.g., Ginges & Atran, 2008;Mummendey, Kessler, Klink, & Mielke, 1999) and advantage and perpetration can evoke heartlessness (e.g., Bandura, 1999;Leach, Snider, & Iyer, 2002), the current results bolster the observation that people who feel that their group has been unfairly disadvantaged or victimized will pursue goals that help them reestablish agency, and people who feel that their group has been unfairly advantaged or perpetrating will pursue goals that help them reestablish communion.
Study 2 additionally examined whether and how (i.e., directly or as a moderator) individuals' dispositions to seek agency and communion in their interpersonal interactions would influence their agentic and communal goals within intergroup interactions.In the case of communion, consistent with Locke (2014), an individual's general disposition to value communion in his/her interpersonal interactions directly predicted the value placed on communion in interactions between the ingroup and other groups.In the case of agency, the ingroup's role in a given context shaped group members' agentic intergroup goals regardless of their preexisting interpersonal dispositions: Even group members who were chronically low in pursuing agentic interaction goals in their interpersonal encounters responded with increased agentic intergroup goals when assigned to the disadvantaged condition.These findings are broadly consistent with research conducted in the intergroup emotions theory framework (Mackie, Devos, & Smith, 2000;Smith, 1993), which revealed a moderate overlap between individual and group emotions but also meaningfully distinct profiles of emotions triggered by the mere contemplation of group membership compared to individual emotion profiles (Seger, Smith, & Mackie, 2009;Smith, Seger, & Mackie, 2007).
Taking both Study 1 and Study 2, involving six different contexts, into account, the overall effects of social role were medium to large by conventional standards (mean effect size iv Cohen's d = 1.14 for agency, d = -0.53 for communion).Thus, in line with the self-categorization perspective (Turner et al., 1987), our results underscore the importance of social context in modulating agentic and communal goals.
A potential limitation of both studies is that we only assessed participants' understanding of their ingroup's rolebut not their identification with their ingroup -within each scenario.Previous research showed that group-based emotions such as collective guilt depend on identification with the respective group (Doosje, Branscombe, Spears, & Manstead, 1998;Maitner, Mackie, & Smith, 2007;Roccas, Klar, & Liviatan, 2006).Thus, future research might examine whether or to what extent ingroup identification moderates the relationship between ingroup role and intergroup goals.
Even though there seems to be growing consensus that the fundamental dimensions of agency and communion can characterize almost every aspect of social interaction (Abele & Wojciszke, 2019), there is quite a debate over which dimension is "the bigger one of the big two" (Abele & Bruckmüller, 2011).By default, when the relationship is harmonious rather than conflictual or threatening, communal information seems to be preferentially processed.
That is, both the self and the ingroup (Leach, Ellemers, & Barreto, 2007) and the other person or group (Abele & Bruckmüller, 2011;Wojciszke & Abele, 2008) will be primarily judged with respect to the "other-profitable" communion dimension.However, research suggests that as soon as there is conflict between individuals or groups, the "weaker" party will experience an increased need to maintain or regain its agency, which brings the "self-profitable" agentic dimension to the foreground (SimanTov-Nachlieli & Shnabel, 2014;SimanTov-Nachlieli et al., 2018).
Consistent with that reasoning, the effects of the ingroup's role were twice as large for agentic interaction goals as for communal interaction goals in the present research.
In conclusion, the results of the present studies provided robust support for the hypothesized association between experiencing the ingroup as illegitimately disadvantaged or victimized and experiencing increased agentic intergroup interaction goals; and it provided some mixed support for the hypothesized association between experiencing the ingroup as illegitimately advantaged or perpetrating transgressions and experiencing increased communal intergroup interaction goals (i.e., the effect of social role on communal goals was not always significant).Understanding the motivational ramifications of self-categorizations as a member of an illegitimately disadvantaged or victimized group or an illegitimately advantaged or perpetrator group is likely to be an important component of any efforts to advance positive intergroup relations and social progress -ranging from structured encounters between members of conflicting groups (i.e., contact interventions), to collective apologies or media campaigns to promote social harmony.

Notes
i) Among German participants, there was a condition (n = 34) originally intended to inform us about group members' baseline agentic and communal intergroup goals (i.e., without assigning them to a particular ingroup role condition).We thought that including this condition would allow to conclude, for example, not only that members of illegitimately disadvantaged/victim groups pursue more agentic goals than members of illegitimately disadvantaged/perpetrator groups, but also that this difference stems from an increase among the disadvantaged/victims (rather than a decrease among the advantaged/perpetrators).In this condition, participants were asked "How should we Germans behave towards other nations?"without referring to a specific outgroup.However, participants in this condition showed significantly lower agentic interaction goals than participants in the illegitimately disadvantaged/victim condition in four of five contexts (ps < .001).Moreover, communion scores in this condition did not differ from communion scores of participants assigned to the illegitimately advantaged/perpetrator role in three of five contexts (international transgression p = .652,exploitative consumption p = .064,gender discrimination p = .317).In hindsight it appears that the chosen context was not truly neutral as intended; possibly due to the history of the Second World War and the unease they might experience with regard to Germany's dominant international role, Germans are prone to conceptualize their nation as perpetrator or illegitimately advantaged in relation to other nations.As such, this condition could not reliably inform us about baseline levels of goals; thus, this condition was dropped from subsequent analyses.
ii) The online survey used the web-based platform "EFS survey", which can adjust the randomization process based on participants' demographical features such as nationality and gender (e.g., whereas participants with migration background were only assigned to contexts 1 and 2, assignment was randomized between these two contexts).

Figure 1 .
Figure 1.Means and 95% confidence intervals of agency and communion vector scores in the different contexts of Study 1 and 2. Note.For the sake of comparability with contexts of Study 1, un-residualized scores are shown for Study 2.

Table 1
Means and Standard Deviations of Perceived Social Role and Effect Sizes of the Social Role Manipulation (Study 1) displays the means and standard deviations of perceived legitimacy in all experimental conditions.Means and Standard Deviations of Perceived Legitimacy of Group Inequality (Study 1) iii Table2

Table 3
Means and Standard Deviations of Agentic and Communal Goals and Effect Sizes of the Social Role Manipulation (Study 1 [Contexts 1-5], and Study 2)

Table 4
Multiple Regression Analyses PredictingAgentic and Communal Interaction Goals From Agentic and Communal Interpersonal Goals, Social  Role, and Legitimacy of Group Inequality (Study 2)