Fast Track Report
Appealing to common humanity increases forgiveness but reduces collective action among victims of historical atrocities
Katharine H. Greenaway, School of Psychology, McElwain Building, The University of Queensland, Brisbane, QLD, 4072, Australia.
Appealing to common humanity is often suggested as a method of uniting victims and perpetrators of historical atrocities. In the present experiment (N = 109), we reveal that this strategy may actually work against victim groups' best interests. Appealing to common humanity (versus intergroup identity) increased forgiveness of perpetrators but independently also served to lower intentions to engage in collective action. Both effects were mediated but not moderated by reduced identification with the victim group. We, thus reveal an important feature of appeals to common humanity: That this strategy may reduce social change at the same time as helping to promote more positive intergroup attitudes. These novel findings extend research on the human identity to a new theoretically interesting and socially important domain. Copyright © 2011 John Wiley & Sons, Ltd.
Human history is full of examples of atrocities perpetrated by one group against another. From historical victimization of Indigenous peoples at the hands of European settlers to modern instances of genocide in Rwanda and Darfur, intergroup conflict seems to be a pervasive element of human interaction. One commonly suggested strategy to mend strained intergroup relations is to appeal to the basic characteristic shared by us all: Our humanity (e.g., United Nations Human Development Report, 2006). After all, if we are all human, we should be able to get along. Although this is a simplified version of the logic, research does show that focusing on common group membership can provide a catalyst for more positive intergroup relations (the common ingroup identity model; e.g., Gaertner, Mann, Murrell, & Dovidio, 1989). Yet, this strategy may have unforeseen consequences in terms of intragroup and intergroup processes.
The social identity perspective (e.g., Tajfel & Turner, 1979; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987) has shown that people can define themselves in terms of different social categories or identities, which have important consequences for how they think and feel about themselves and others. According to self-categorization theory, people categorize themselves at three levels of increasing inclusiveness: personal, social, and human. When individuals shift from categorizing at the social level to a more inclusive superordinate categorization, previous outgroup members are accepted as part of the shared ingroup, and thus elicit more positive evaluations (e.g., Dovidio, Gaertner, & Saguy, 2009; Eller & Abrams, 2004; Reicher, Cassidy, Wolpert, Hopkins, & Levine, 2006). Yet, despite the utility of humanity as an overarching category within which all possible groups are subsumed, comparatively little empirical research has investigated the effects of categorization at the ultimate human level.
Recent research has shown that human categorization may not have the universally positive effects expected among perpetrator group members (e.g., Greenaway & Louis, 2010; Morton & Postmes, 2011). However, Wohl and Branscombe (2005) found that a focus on shared humanity improved intergroup attitudes among representatives of victimized groups (e.g., Jewish and Native Canadians). Specifically, they found that victim group members who categorized perpetrators as fellow humans were more likely to forgive and less likely to assign collective guilt for past harms compared with people for whom the subgroup victim identity was made salient. Promoting more positive feelings towards perpetrator groups is an important step in repairing damaged intergroup relations, and forgiveness is arguably a critical step in achieving intergroup reconciliation (e.g., Karremans & van Lange, 2004; Noor, Brown, Gonzalez, Manzi, & Lewis, 2008 but cf., Hamber, 2007). Clearly, human categorization can promote positive intergroup attitudes among victim group members, but recent theorizing suggests that this may not be the only consequence of a focus on common humanity.
Wright and Lubensky (2009) recently presented a theoretical framework to unify separate literatures on prejudice reduction and collective action. Their analysis revealed an important theoretical paradox that interventions aimed at promoting positive intergroup relations may improve attitudes among disadvantaged group members but simultaneously reduce their motivation to engage in collective action to redress social inequalities. The common ingroup method was particularly implicated as a strategy that may have these unintended consequences. The present research investigates this logic at the ultimate human level and identifies a novel but theoretically grounded process through which human categorization might influence intergroup attitudes: Lowered subgroup identification.
THE ROLE OF IDENTIFICATION
A focus on common humanity as an inclusive superordinate identity could undermine the salience of subgroup identity. Indeed, shifting identification from the subgroup to superordinate level is the process through which intergroup hostility is hypothesized to be reduced in the common ingroup method (e.g., Dovidio et al., 2009). This may be particularly likely among victim group members who have experienced and sometimes continue to experience disadvantage and thus have much to gain from inclusion in a comparatively high-status superordinate category (Hornsey & Hogg, 2002). As such, lowered subgroup identification may act as a mediating process through which human categorization influences intergroup attitudes.
It is well known that group identification influences people's attitudes and behavior. For example, much research has shown that collective action is most likely when people are highly identified with their ingroup (e.g., Simon, Trötschel, & Dähne, 2008; Wright, 2001). Based on this reasoning, it is logical to expect that if victims are less identified with their own group in favor of common human identity, they may be less likely to engage in collective action. This process of reduced subgroup identification may also account for the change in intergroup attitudes found in previous work on human categorization, such as increased forgiveness or lowered collective guilt assignment (Wohl & Branscombe, 2005). The present research tests this novel logic. We propose that by reducing the salience of subgroup identification, human categorization may improve intergroup attitudes among victim group members at the same time as relaxing their motivation to achieve social change.
Another possibility is that subgroup identification will act as a moderator of the effects of human categorization. Extensive research in the social identity tradition reveals that high identifiers—those for whom the group is of central importance—act differently to low identifiers. For example, high identifiers have been shown to feel less guilty for ingroup-perpetrated harm (Doosje, Branscombe, Spears, & Manstead, 1998) and are less forgiving of harm perpetrated against the ingroup than low identifiers (Brown, Wohl, & Exline, 2008). If people attach great importance to membership in a particular group, they may be less willing to let go of this identity in favor of a higher-order one (Hornsey & Hogg, 2000). As such, appealing to common humanity may only succeed in influencing intergroup attitudes among low identifiers for whom the subgroup identity is less important to begin with.
The present research investigates the as yet untested role of identification in appeals to common humanity. Based on theorizing by Wright and Lubensky (2009) and consistent with the principles of the common ingroup identity model, we hypothesized that subgroup identification would act as a mediator, rather than a moderator, of the effects of human categorization. In particular, we expected that human categorization would reduce collective action intentions at the same time as improving intergroup attitudes among victim group members through the process of lowered subgroup identification. The present research thus investigates a possible mechanism underlying the effects of human categorization on intergroup attitudes and reveals hidden depths to the generally positive anticipated effects of appealing to common humanity.
The experiment was conducted in the Australian context. Participants were 109 (82 female) self-identified Indigenous Australians who volunteered to complete the study. The majority of participants (n = 80) completed the study online and were recruited from various Indigenous networks. The remainder (n = 29) were approached at urban Indigenous community centers and invited to participate. The procedure of data collection was identical across the two samples, which were pooled for analysis. Random assignment was successful; there were no age (t(108) = 0.01, p = .887) or gender differences (χ²(109) = 1.42, p = .262) between the experimental conditions. Participants ranged in age from 17 to 65 with a mean age of 36.84 (SD = 11.40).
The experiment employed a one-way ANOVA between subjects design in which categorization (human versus intergroup) was manipulated. The dependent variables were forgiveness of perpetrators, collective guilt assignment, and intentions to engage in collective action. The mediating and moderating effects of subgroup identification were tested.
Procedure and Measures
Participants were randomly assigned to one of two categorization conditions making either the human or intergroup identity salient. Following Wohl and Branscombe (2005), participants were asked to think about a historical atrocity perpetrated against their group. Participants reflected on the events of the Stolen Generations, which took place in Australia during the 20th Century, in which Indigenous children were forcibly removed from their families (Human Rights and Equal Opportunity Commission, 1997). In the human categorization condition (n = 54), the events were described as examples of humans behaving heartlessly and ignorantly towards other humans. In the intergroup categorization condition (n = 55), the events were described as White Australians behaving heartlessly and ignorantly towards Indigenous Australians. To reinforce the intergroup categorization, participants in this condition indicated their ethnicity immediately following the manipulation. Participants in the human categorization condition indicated their ethnicity at the end of the questionnaire.
All questions were scored using a Likert-type scale ranging from 1 (strongly disagree) to 8 (strongly agree). Success of the categorization manipulation was assessed using a single-item measure from Wohl and Branscombe (2005; “Indigenous Australians and White Australians share basic similarities”). Four items, adapted from Hogg and Hains (1996) measured subgroup identification (e.g., “I feel similar to other Indigenous Australians”, α = .72). Forgiveness of perpetrators was measured using four items adapted from Wohl and Branscombe (e.g., “White Australians today should be forgiven for what their group did to Indigenous Australians during the years of the Stolen Generations”, α = .86).
Two items adapted from Wohl and Branscombe measured collective guilt assignment (“White Australians of today should feel guilty about the bad things that happened to Indigenous Australians during the years of the Stolen Generations” and “Today's White Australians should feel guilty about the awful things their ancestors did to Indigenous Australians during the years of the Stolen Generations”, α = .96). Three items adapted from Lalonde and Cameron (1993) measured willingness to engage in collective action on behalf of the victim group (“I would participate in a peaceful demonstration aimed at bettering the position of Indigenous people in Australia”, “I would lobby the Government to improve the position of Indigenous people in Australia”, and “I would give up some of my personal time in order to help out people in the Indigenous community”, α = .67).
The categorization manipulation was successful, F(1,107) = 10.23, p = .002, ηp² = 0.088. Participants in the human categorization condition perceived victims and perpetrators as more similar (M = 4.57, SD = 2.83) than participants in the intergroup categorization condition (M = 3.02, SD = 2.21).
The means and standard deviations from the experimental effects are displayed in Table 1. All effects were in line with the hypotheses. A series of one-way ANOVAs revealed that participants in the human categorization condition were significantly less identified with the subgroup than participants in the intergroup categorization condition, F(1,106) = 13.31, p < .001, ηp² = 0.112. In addition, victim group members who were induced to categorize at the human level were more forgiving of perpetrators, F(1,106) = 7.65, p = .007, ηp² = 0.067, assigned marginally less collective guilt, F(1,107) = 3.03, p = .085, ηp² = 0.027, and reported lower collective action intentions, F(1,106) = 8.23, p = .005, ηp² = 0.073, than participants in the intergroup categorization condition.
Table 1. Means and standard deviations of victim group identification, forgiveness, collective guilt assignment, and collective action intentions by categorization condition
|Subgroup identification||6.99 (1.06)||7.60 (0.62)|
|Forgiveness of perpetrators||5.02 (2.00)||3.98 (1.93)|
|Collective guilt assignment||4.40 (2.51)||5.21 (2.36)|
|Collective action intentions||6.75 (1.37)||7.40 (0.94)|
Mediational analyses were performed to determine whether the effects of human categorization on the dependent variables could be explained by reduced identification with the victim group. Because we know from the ANOVAs that level of categorization had direct effects on our variables of interest, we proceeded to test the model in which the dependent variables were regressed onto the categorization manipulation (coded as 1, human and −1, intergroup) and the mediator, subgroup identification. Following Preacher and Hayes (2008), the indirect effects of human categorization on the dependent variables were computed from unstandardized regression weights with 5000 bootstrap resamples. Intercorrelations among the variables are displayed in Table 2.
Table 2. Means, standard deviations, and bivariate correlations between the categorization manipulation and dependant variables
|1. Human categorization manipulation|| ||−.33***||.26**||−.17†||−.27**|
|2. Subgroup identification|| || ||−.43***||.24*||.41***|
|3. Forgiveness of perpetrators|| || || ||−.27**||−.20*|
|4. Collective guilt assignment|| || || || ||.19*|
|5. Collective action intentions|| || || || || |
Categorization and subgroup identification accounted for a significant amount of variance in forgiveness, R² = 0.20, F(2,104) = 13.10, p < .001. Greater subgroup identification was associated with less forgiveness of perpetrators, β = −.37, p < .001, and with the mediator included in the model the categorization manipulation was no longer a significant predictor, β = .15, p = .105. Bootstrapping analyses confirmed that the indirect effect of human categorization on forgiveness via reduced subgroup identification was significant (IE = 0.26, SE = 0.09, 95% CI = 0.11,0.47).
Collective Guilt Assignment
Categorization and subgroup identification accounted for a significant amount of variance in collective guilt assignment, R² = 0.07, F(2,104) = 3.83, p = .025. Greater subgroup identification was associated with more guilt assignment, β = .23, p = .023, and with the mediator included in the model the categorization manipulation dropped from marginality to non significance, β = −.06, p = .531. The indirect effect of human categorization on guilt assignment via reduced subgroup identification was significant (IE = −0.17, SE = 0.09, 95% CI = −0.42,−0.01).
Categorization and subgroup identification accounted for a significant amount of variance in collective action intentions, R² = 0.19, F(2,104) = 12.22, p < .001. Greater subgroup identification was associated with intentions to engage in collective action, β = .37, p < .001, and with the mediator included in the model the categorization manipulation was no longer a significant predictor, β = −.13, p = .110. The indirect effect of human categorization on collective action intentions via reduced subgroup identification was significant (IE = −0.14, SE = 0.05, 95% CI = −0.33,−0.04).
In addition to subgroup identification, we tested forgiveness as an alternative mediator of the effects of human categorization on collective action intentions. Forgiveness did not predict collective action intentions, β = −.26, p = .175, and the effect of the categorization manipulation remained significant with the alternative mediator in the model, β = −.22, p = .024. The indirect effect of human categorization on collective action intentions via increased forgiveness was not significant (IE = −0.05, SE = 0.04, 95% CI = −0.14,0.02).
Moderated multiple regression analyses were conducted to test the possibility that subgroup identification acted as a moderator as well as a mediator of the effects of human categorization. The categorization manipulation and centered subgroup identification variable were entered in Block 1, followed by the interaction in Block 2. The interaction between categorization and identification failed to predict variance in forgiveness, β = .04, p = .680, collective guilt assignment, β = .10, p = .382, and collective action intentions, β = .03, p = .787.
The importance of a uniting superordinate identity is at the heart of many contemporary and classic theories of intergroup relations. The present research adds to the burgeoning body of work investigating the ultimate superordinate category, humanity, and reveals new and unexpected effects of appealing to this identity. Consistent with previous work (Wohl & Branscombe, 2005), victim group members reported greater forgiveness of perpetrators and lower collective guilt assignment when categorized at the human compared with the intergroup level. The present research adds another layer of complexity to these effects, however, by revealing reduced collective action as a paradoxical consequence of human categorization, and lowered subgroup identification as a novel mediator through which appeals to common humanity influence victim attitudes and intentions.
It seems intuitive that focusing on a superordinate identity would reduce the salience of subgroup identity, yet the present research represents one of the only tests of this logic and is the first test at the maximally inclusive human level. Importantly, we tested identification as both a potential mediator and moderator of the effects of human categorization. It is possible to imagine based on research in the social identity tradition that identification might play either—or both—roles. Consistent with the underlying principles of the common ingroup method, our analyses revealed that subgroup identification acted as a mediator rather than a moderator of the effects of appealing to common humanity.
The mediating effect of reduced subgroup identification on forgiveness is consistent with recent correlational research showing that less identified victim group members forgive perpetrators more readily (e.g., Brown et al., 2008, Study 2). From these effects, we might conclude simply that appealing to common humanity improved intergroup attitudes among victims in line with previous interpretations (Wohl & Branscombe, 2005). For reconciliation to occur, victim group members need to let go of negative attitudes and feelings toward perpetrators (e.g., Noor et al., 2008) and thus human categorization may be successful in promoting more positive intergroup relations. A change in attitudes was not the only consequence of lowered subgroup identification resulting from human categorization, however.
Despite the modest effect size, an experimentally induced focus on common humanity significantly depressed intentions to engage in collective action. As expected, this novel effect was mediated by reduced identification as a victim group member. Identification with a group is a critical prerequisite for people to consider engaging in social change (e.g., Wright, 2001; see also Louis, 2009). Reduced collective action is therefore a logical outcome of strategies that focus attention away from subgroup membership and onto common humanity.
We also tested forgiveness as an alternative mediator of the effects of human categorization on collective action intentions. It is plausible to imagine that having forgiven perpetrators victim group members may no longer feel a need to protest against them. From this perspective, by encouraging victims to let go past grievances, appeals to common humanity may reduce the perceived need for militant group-based action in the present. Our analysis of forgiveness as a potential mediator of the effects of human categorization rules out this alternative explanation. Instead, appealing to common humanity reduced subgroup identification, which independently promoted forgiveness of perpetrators and reduced intentions to engage in collective action among victim group members.
The present research sheds new light on the effects of appealing to the ultimate common ingroup—humanity. We reveal that appeals to common humanity have the potential to lay to rest grievances for wrongdoing in the past, at the same time as reducing the motivation to seek social change in the present. According to our research, therefore, the consequences of an appeal to common humanity are mixed from the victim's perspective. Forgiveness has been linked with increases in psychological well-being among victim group members (e.g., Bono, McCullough, & Root, 2008), suggesting the possibility for positive intragroup as well as intergroup, effects from a focus on shared humanity. On the other hand, by reducing intentions to engage in collective action, this strategy may undermine victim groups' efforts to achieve social change for their group. We suggest, therefore, that appeals to common humanity might sometimes work against victims' best interests: Promoting benignly positive intergroup attitudes while depressing the motivation to achieve true social change.