The impact of including advantaged groups in collective action against social inequality on politicized identification of observers from disadvantaged and advantaged groups

Mobilizing public support is key to a movement’s success. Little is known, however, about how movements can achieve this goal and whether involving advantaged group members is beneficial for a movement’s cause. In a set of five experiments with convenience samples collected in the United States and Germany (total N = 1,625), we examined whether protests (e.g., against racism and sexism) with and without advantaged group members affect politicized identification among observers. We expected that the presence (vs. absence) of advantaged group members at a protest will increase politicized identification among advantaged group observers, which was confirmed in Studies 1A– 1C. In contrast, we expected that the disadvantaged group observers will increase or decrease their politicized identification depending on the role advantaged group members have at a protest (i.e., supportive vs. leadership role). Studies 2A– 2B revealed that when advantaged group members had a supportive role, disadvantaged and advantaged group observers increased their politicized identification, but this effect was absent when they had a leadership role. Moreover, including advantaged group members in a protest increased the belief that solidarity is a normative behavior and the expectations that a protest will be peaceful among observers. Implications for research on allyship are discussed.

. Public response to this campaign has been mixed: The campaign received praise, but it was also seen as controversial and criticized for its portrayal of men as 'saviors' of women (McCarthy, 2014).Similarly, the involvement of White people in the antiracism protests, sparked by the police killing of George Floyd in the United States, has been met with enthusiasm and suspicion about their true intentions and commitment to the movement (Parker, 2020).
This article seeks to understand when and why societal reactions to the advantaged group's participation in collective action against inequality may be positive (or negative).We examine how the inclusion of advantaged group members affects observers' politicized identification-the strongest psychological predictor of engagement in collective action (Stürmer & Simon, 2004;Van Zomeren, Postmes, & Spears, 2008).Social-psychological theorizing suggests that a movement's success in challenging the status quo is contingent on its power to create a shared identity with the general public in order to mobilize them for its cause (Simon & Klandermans, 2001).Hence, the inclusion of advantaged group members in collective action may be valuable, because it can help movements reach a broader audience necessary to achieve social change (Louis, 2009;Subašić, Reynolds, & Turner, 2008).At the same time, members of the general public who belong to the disadvantaged group may perceive the presence of advantaged group members as hijacking their struggle (Droogendyk, Wright, Lubensky, & Louis, 2016), which may decrease their politicized identification.We review the literature that supports both of these conclusions before presenting our studies.

Collective Action and Politicized Identification
Collective action is defined as any action undertaken by individuals on behalf of a group to achieve the group's goals (Wright, Taylor, & Moghaddam, 1990).Typically, the majority of socialpsychological research focuses on the motivations of disadvantaged group members to engage in collective action.The key motivation for collective action is individuals' identification with the group and especially its politicized form, that is, identification with a social movement organization or an activist group (Van Zomeren et al., 2008).This is because politicized identification reflects the extent to which individuals have internalized the collective grievances as their own and their readiness to engage in the political arena to fight for the group's goals (Simon & Klandermans, 2001).For instance, previous research found that gay men's identification with the German gay movement, an organization fighting for the rights of gay men, predicted participation in collective action whereas identification with a broader category of gay men did not (Stürmer & Simon, 2004).
Moreover, politicized identity is central to understanding how the general public becomes sympathetic to a movement's cause.Simon and Klandermans (2001) argue that activists do not only engage in collective action against the adversaries, but they also seek to attract third-party support for their cause.Social-psychological theorizing on political solidarity suggests that social movements need to create a sense of common cause and shared identity with the broader society to be able to gain their support (Subašić et al., 2008).Politicized identity represents the key psychological bond between the movement and the general public: It captures the public's support for the movement' struggle, as well as opposition to the status quo and those in power.Previous research found that politicized identification predicts engagement in solidarity-based actions among advantaged group observers (Van Zomeren, Postmes, Spears, & Bettache, 2011).
Less is known, however, about how movements can strengthen the psychological bond with their audience.Activists may communicate their values, severity and urgency of the issue, as well as the movement's efficacy in securing its goals to motivate potential followers (Benford, 1993).Moreover, the research on opinion-based groups shows that interactions with like-minded others facilitate politicized identification and commitment to action (e.g., Thomas & McGarty, 2009).In this article, we examine a novel factor and propose that movements can secure public support by involving advantaged group members in collective action, but we also highlight the potential costs associated with their involvement.

The Consequences of Involving Advantaged Group Members in Collective Action Against Social Inequality
Advantaged group members can be a valuable asset to a movement fighting against inequality, because they belong to the group that has power and can use their privilege and status to challenge the powerholders and secure public support for a movement's cause (Subašić et al., 2008).According to resource mobilization theory, advantaged group allies can be seen as conscience constituents (McCarthy & Zald, 1977) who contribute to a movement not because they seek to benefit from collective action, but presumably out of moral values and beliefs (though other work argues that advantaged group allies may also act out of self-interest; see Radke, Kutlaca, Siem, Wright, & Becker, 2020).Thus, allies may exert a positive influence on the observers because they are perceived as role models and moral exemplars.For instance, advantaged group allies were more likely to change the attitudes and opinions among conservative members of their group (Maas, Clark, & Haberkorn, 1982) and reduce discriminatory behaviors among advantaged group members (Czopp & Monteith, 2003).Likewise, Subašić and colleagues (2018) found men to be more supportive of gender-equalizing policies at work when they read an appeal made by a male compared to a female leader of a gender-equality group.Louis (2009) suggested that actions taken by advantaged group allies may expose how the inequality violates the norms and values of the advantaged group itself, which may motivate other advantaged group members to decrease their prejudicial behavior by conforming to those who openly condemn it.Norms are beliefs shared by group members that describe what they do (i.e., descriptive norms) or should ideally do (i.e., injunctive norms) in a given situation (Terry & Hogg, 1996).Importantly, norms exert influence on individual's behavior when they are made salient (Kallgren, Reno, & Cialdini, 2000) and are psychologically relevant (Smith & Louis, 2008).Because advantaged group members are more likely to be seen as normative and prototypical (Wenzel, Mummendey, & Waldzus, 2008), their involvement in collective action may signal to the general public that supporting the disadvantaged group's fight has become the new norm.Consequently, members of the general public may increase their politicized identification by conforming to this norm.
Additionally, the presence of advantaged group members may affect public expectations about whether the protest will be peaceful or violent.Often, protests tend to disrupt daily routines, and the exposure to protests can have a negative impact on bystanders' support for a social movement (Selvanathan & Lickel, 2019).Moreover, protests that result in the destruction of public property and clashes with police transgress societal norms and are more likely to be perceived negatively by the public irrespective of their cause (Feinberg, Willer, & Kovacheff, 2020).Feinberg and colleagues (2020) found that violent and extreme protest tactics reduced public support for social movements by decreasing their identification with the movement.However, actions by advantaged group members, such as confrontation of prejudice, are less likely to be seen as aggressive and are more likely to be seen as legitimate than the same actions enacted by disadvantaged group members (Czopp & Monteith, 2003).Additionally, police are less likely to use force against advantaged group members compared to disadvantaged group members (Kahn, Goff, Lee, & Motamed, 2016).Thus, advantaged group's presence may have a positive impact on the public's impressions of the protest: The general public may perceive a protest including advantaged group members as more likely to be peaceful, which in turn may increase their politicized identification.
However, involving advantaged group members may also be costly for a movement.Research on intergroup helping finds that advantaged group members are sometimes motivated to help out of selfish concerns, such as to protect their personal image and/or maintain the status of their own group (Halabi, Dovidio, & Nadler, 2008).Not surprisingly, disadvantaged group members are less likely to accept such help (Nadler & Halabi, 2006).Furthermore, if disadvantaged group members perceive that the focus has shifted too far too accommodating advantaged group members in an effort to achieve harmonious intergroup relations, they are less willing to participate in collective action (Dixon, Levine, Reicher, & Durrheim, 2012;Hasan-Aslih, Pliskin, van Zomeren, Halperin, & Saguy, 2019).Droogendyk and colleagues (2016) suggested that advantaged group activists can undermine the movement by insisting on their voices being heard instead of providing support and seeking guidance from disadvantaged group activists.Importantly, Iyer and Achia (2020) found that disadvantaged group members are less likely to support social-justice organizations led by advantaged groups than by disadvantaged groups.This means that disadvantaged group observers may be suspicious of advantaged groups' true intentions and interpret their presence at a protest as wanting to take the center stage, which may decrease their politicized identification.In line with Iyer's and Achia's findings (2020), this is more likely to happen when advantaged group members assume a leading role in the protest then when they stay in the background (see Figure 1 for the overview of the model).

Summary and Overview
We examine the effect of including advantaged group members (vs.not including them) on observers' politicized identification in five experimental studies across two different contexts.We presented members of the general public who belong to disadvantaged and advantaged groups with short descriptions of an upcoming protest against social inequality, which will be attended by either members of the disadvantaged group only or by members of disadvantaged and advantaged groups together.In line with previous social-psychological research (Van Zomeren et al., 2008), we operationalized politicized identification as identification with the social movement and/or the protesting group.
Based on the literature reviewed above, we formulated two hypotheses.First, in line with the literature showing positive effects of involving advantaged group allies, we expected that H1: The presence of advantaged group members at a protest will increase politicized identification among the observers from the advantaged group.
Second, we expected that the reactions of disadvantaged group observers will depend on how the advantaged group's presence at the protest is perceived (Droogendyk et al., 2016).Specifically, we tested whether H2A: The observers from the disadvantaged group may increase their identification if they perceive the advantaged group members as supportive; H2B: The observers from the disadvantaged group may decrease their identification if they perceive the advantaged group members as taking over the protest.
In Study 1A-1C, we varied the percentage of advantaged group members who were planning to attend a protest (25% vs. 50%), and we reasoned that the larger presence should strengthen the proposed effects.In Study 1C, we further explored whether the expected effects on politicized identification result from the changes in norm perceptions (i.e., norms about the support for protests against social inequality) and the expectation that the protest will be peaceful.In Study 2A-2B, we manipulated whether the advantaged group members have a supportive or a leading role at a protest to further examine the reactions of observers from the disadvantaged group.Additionally, we tested the mediating role of norm perceptions and nonviolence expectations.

STUDY 1A AND STUDY 1B
Study 1A and Study 1B examined whether the presence of advantaged group members has an impact on observers' politicized identification in two different contexts: collective action against the discrimination of Black Americans in the United States and collective action against the discrimination of women in Germany.These two studies were exploratory, and we included a broad range of questions about public perceptions of collective action, allyship, stereotypes of activists, etc., which are not of focal interest in this article.All studies reported in this article were approved by the ethics committee of the University of Osnabrück.Complete questionnaires, databases, and the syntax can be found at https://osf.io/p3dc5/

Study 1A Participants
Five hundred and fourteen Amazon Mechanical Turk workers participated in the study for a small monetary reward of $0.50.The data was collected in December, 2018.Following the recommendations by Hauser and colleagues (2019), we included an attention check specific to our manipulation.This resulted in 133 responses being excluded from the analyses: 74 who failed the attention check, 34 who did not identify as Black or White Americans, and 25 who did not fill out the survey seriously.More details about the samples for each study including exclusion criteria and power analyses can be found in the Appendix S1 in the online supporting information.The responses of 381 individuals (204 Black Americans, 177 White Americans; M age = 37.67, SD = 11.66;58.3% women) were included in the final analyses.The study was administered in English language via the Qualtrics survey platform.

Study 1B Participants
Two research assistants recruited 268 individuals living in Germany to participate in the study for the possibility of winning a €20 Amazon voucher.The data was collected in March and April, 2018.We excluded 59 responses from the analyses: 50 who failed the attention check, 7 who were younger than 18 at the time of the survey, and 2 who identified as gender non binary.The final sample used in the analyses consisted of 209 participants (109 women and 100 men; M age = 33.53,SD = 13.53).The study was administered in German language via the Unipark survey platform.

Manipulation
In both studies, the participants read a short newspaper excerpt about an upcoming protest organized by the Black Lives Matter Movement against police killings of Black Americans in New York (Study 1A), or an upcoming protest in Berlin, which demanded equal representation of women in leadership positions (Study 1B).We manipulated the expected composition of the protesters: In the control condition, the participants read that only members of the disadvantaged group (i.e., Black Americans/women) were planning to attend the protest, whereas in the two advantaged group conditions they read that 25% or 50% of the people planning to attend the protest were members of the advantaged group (i.e., White Americans/men).We also instructed the participants to take a minute and imagine how the protest might look like.The exact wording of the manipulation and the items for each study can be found in the Appendix S2 in the online supporting information.

Politicized Identification
In Study 1A, the participants expressed their agreement (1 = Strongly disagree to 7 = Strongly agree) with two items asking whether they identified with and perceived a shared common background with other supporters of the Black Lives Matter movement (Spearman-Brown reliability coefficient = .92).
In Study 1B, we changed to identification with a protesting group instead of a known movement, as a better proxy of politicized identification.We reasoned that identification with a known social movement may be less malleable to contextual influences.Thus, we asked the participants whether they identified with and perceived a shared common background with protesters who were planning to attend the event (Spearman-Brown reliability coefficient = .92).Including Advantaged Groups in Collective Action

Attention Check
At the end of the survey, we asked the participants whether they remembered who was going to participate in the protest.The participants could choose one of the four options: (1) only disadvantaged group members versus (2) 75% disadvantaged and 25% advantaged group members versus (3) 50% disadvantaged and 50% advantaged group members (4) I don't remember.Below we report the analyses only on the participants who answered the question correctly.

Study 1A Results
We ran a univariate analysis of variance with Protest Manipulation (Control vs. 25% Advantaged group members vs. 50% Advantaged group members) and Audience (Disadvantaged group vs. Advantaged group) on politicized identification.The analysis on politicized identification yielded a statistically significant main effect of Audience, F(1,375) = 97.00,p < .001,η p 2 = .21,no significant main effects of Protest Manipulation F(2,375) = .27,p = .761,but a significant interaction effect, F(2,375) = 3.57, p = .029,η p 2 = .02.Not surprisingly, disadvantaged group observers identified more strongly as supporters of the movement (M = 5.20, SE = .14)than advantaged group observers (M = 3.20, SE = .15).Next, we followed up a significant interaction effect by running simple maineffect analyses for each group using Bonferroni correction.However, we did not find statistically significant differences between the control and ally conditions for either disadvantaged or advantaged group observers (all ps > .1).The pattern of means suggested that advantaged group observers identified somewhat more strongly as supporters of the Black Lives Matter movement the more advantaged group members were expected to be present at the protest, whereas disadvantaged group observers identified somewhat less strongly (see Figure 2 for more details).
The results on politicized identification provide initial support for Hypothesis 1 that the presence of advantaged group members increases politicized identification among advantaged group observers, at least when the expected ratio of disadvantaged and advantaged group protesters was about equal.The effects on the observers from the disadvantaged group remained unclear.In line with previous work on the importance of group norms (Louis, 2009), the presence of advantaged group members may be more psychologically relevant for advantaged compared to disadvantaged group observers.However, we did not measure norm perceptions nor expectations that the protest will be peaceful in Study 1A-1B.We address this limitation in Study 1C.

Participants
Five hundred and seven Amazon Mechanical Turk workers participated in the study for a small monetary reward of $0.50, but we excluded 183 responses from the analyses: 85 who failed the manipulation attention checks, 38 did not identify as Black or White American, and 60 who did not fill  out the survey seriously (for more details, see Appendix S1 in the online supporting information).The data was collected at the beginning of July, 2018.The final sample consisted of responses of 324 individuals (169 Black Americans, 155 White Americans; M age = 37.35, SD = 12.35; 50.3% women).The study was administered via the Qualtrics survey platform.

Manipulation
The manipulation was similar to the one used in Study 1A, except that it referred to an upcoming protest against police killings of Black Americans in Chicago.Moreover, we emphasized that the protest is supported and will be attended by the local community, that is, Black Americans or Black and White Americans living in Chicago depending on the condition.Additionally, the participants were asked to write down their thoughts about the protest to increase the strength of the manipulation.

Politicized Identification
The participants responded to two items asking whether they share a common ground and identify with people planning to attend the protests (Spearman-Brown reliability coefficient = .92).

Norm Perceptions
We operationalized the norm perceptions as the beliefs about advantaged and disadvantaged group members' support for protests against racism in general.The participants were instructed to reflect about whether other people would support the protest similar to the one they read about.They then responded to four items asking separately whether other Black or White Americans are supportive of these protests (i.e., descriptive norm) and think they should engage in protests against police violence (i.e., injunctive norm).We combined two items measuring perceived norms about the engagement of White Americans and two items measuring the perceived norms about the engagement of Black Americans (Spearman-Brown reliability coefficient = .88and .90respectively).

Peaceful Protest Expectations
The participants responded to six items asking how likely it is that the protest would be peaceful and safe or violent and dangerous.We recoded the violent items, so higher scores on the scale indicate more agreement that the protest would be peaceful.Exploratory factor analysis with Oblimin rotation extracted one factor with Eigen value larger than one explaining 65.30% variance.We created the scale using the original items (α = .92).

Attention Check
We used the same attention check as in Study 1A.
Next, we found a significant main effect of Protest Manipulation on norm perceptions regarding the engagement of White Americans, F(2, 318) = 3.63, p = .028,η p 2 = .02.There were no other significant effects on this variable.The manipulation strengthened the perceived norm regarding the support of advantaged group members, but only when the participants were told that 50% of protesters would be White Americans in contrast to the control condition: M 50% = 4.49, SE = .16vs. M control = 3.87, SE = .17,p = .022,95% CI [.07, 1.18].There were no significant differences between the control and 25% condition: M 25% = 4.19, SE = .15vs. M control = 3.87, SE = .17,p = .461,95% CI [−.22, .86].No other effects were significant.
Second, there were no significant effects of norm perceptions regarding the engagement of Black Americans.Disadvantaged and advantaged group observers believed that Black Americans are generally very supportive of the protests against police violence (M = 5.35, SD = 1.54), irrespective of who was planning to attend the protest.

Discussion
Study 1C provided further support for Hypothesis 1-that a protest including 50% allies increases politicized identification among advantaged group observers.Importantly, we identified two possible reasons for this effect: The anticipated presence of advantaged group members increased the perceived norms that the advantaged group is generally becoming more supportive of a protest against racism and the expectations that the protest will remain peaceful.However, as in previous studies, we observed this effect only when the ratio between disadvantaged and advantaged group protesters was equal.Focus theory of normative behavior suggests that increasing the salience of a given norm will increase its influence on individual behavior (Kallgren et al., 2000).It is possible that the 25% condition, in contrast to the 50% condition, was not strong enough to make the norm about the advantaged group's support for the antiracism protest salient.
On the other hand, the findings for disadvantaged group observers remain inconclusive.In contrast to the first two studies, disadvantaged group observers increased their politicized identification.The manipulation in Study 1C emphasized the advantaged group's support for the protest, which may have resonated well with disadvantaged group observers.One reason for the inconsistent results across the three studies should derive from the general lack of information and clarity regarding the role of advantaged group members at the protest: We only communicated that they would attend the protest, but we did not specify what they would be doing there.We address this limitation in the next study.

STUDY 2A-2B
Previous research suggests that disadvantaged group members are sensitive to the role advantaged group members play in the organizations aimed to benefit their group.For instance, Iyer and Achia (2020) found that when a leadership team consisted of more advantaged than disadvantaged group members rather than the opposite, disadvantaged groups reported less willingness to act with such an organization.In contrast, advantaged groups' intentions were not affected by the composition of a leadership team.We expected similar processes at play when it comes to the presence of advantaged group members at protests.To test this, we made several important changes to our design: We dropped the 25% condition and included two variations of the 50% condition-one in which the advantaged group members were there to support the protest and one in which they would lead the protest.We expected that reading about a protest where advantaged group members have a supportive role will increase politicized identification among disadvantaged group observers in line with Hypothesis 2A.In contrast, we expected that reading about a protest where advantaged group members take a leadership role will decrease politicized identification among disadvantaged group observers in line with Hypothesis 2B.We assumed that the role advantaged group members have at a protest would not matter to advantaged group observers.Lastly, we included a mediation analysis to test whether the proposed mechanisms (norms and expectations about peaceful actions) mediated the effect of our protest manipulation on politicized identification.
In order to increase the generalizability of our findings, we conducted the same study in two different protest contexts: protests against racism and sexism in the United States.The studies were run in the same cultural context and used the same items, therefore we opted for an integrative dataanalysis approach (Cumming, 2014).Namely, we ran the analysis on the pooled data and controlled for the effects of context, as suggested by Curran and Hussong (2009).Some of the key advantages of an integrative data-analysis approach are increased statistical power and larger heterogeneity of the sample, which were important to us as the key effects in our previous set of experiments were small.The results on the separate datasets replicate the findings from the pooled analyses (see Appendix S4 in the online supporting information).

Participants
We recruited 1,260 participants on Amazon Mechanical Turk and paid them $0.50 for their participation.Six hundred and forty-five individuals took part in the study on the protest against police brutality against Black Americans (Study 2A) and 615 in the study on the protest against discrimination of women in the workplace (Study 2B).The two datasets were collected in September 2019, one week apart, and the participants could only take part in one study.After screening the data, we excluded 549 responses: 247 who failed the attention checks, 58 responses which came from identical IP addresses, 202 who failed to complete the survey or provided uninterpretable responses, 38 who did not identify as Black/White American (in Study 2A), and four individuals who identified as gender nonbinary or did not fill out the question (in Study 2B).More details can be found in the Appendix S1 in the online supporting information.The final sample in Study 2A consisted of 140 Black Americans and 184 White Americans (M age = 37.56, SD = 12.22; 59.9% women).The final sample in Study 2B consisted of 217 women and 170 men (M age = 37.10, SD = 11.74;79.6% White Americans).The studies were administered in English language via the Qualtrics survey platform.

Manipulation
The manipulations used in both studies were exactly the same except one described an upcoming protest against police brutality against Black Americans and the other described an upcoming protest against discrimination of women in the workplace.In order to enhance the credibility of the manipulations, we presented them as Facebook posts by a local newspaper publisher in San Francisco.The participants were randomly assigned to one of the three conditions: control versus advantaged group-supportive role versus advantaged group-leadership role.In the control condition, the participants read that only disadvantaged group members will be attending the protest (like in Studies 1A-1C), and they were also informed that famous disadvantaged group activists (Black Americans/women) will be addressing the crowd and leading the protest.In the advantaged group-supportive-role condition, the participants read that half of the people attending the protest will be advantaged group members, and famous disadvantaged group activists would be leading the protest.In advantaged group-leadership-role condition, in contrast to the other two conditions, the participants read that four well-known activists from the advantaged group (i.e., White American activists/male activists) and only one activist from the disadvantaged group would be leading the protest.

Dependent Variables
The participants filled out a four-item measure of politicized identification: In addition to the two items we had in previous studies, we asked the participants to what extent they felt solidarity and a bond with the protesters (α Study2A = .96and α Study2B = .97).Norms and protest expectations were assessed with the same items as in Study 1C.Reliability analyses can be found in the Appendix S4 in the online supporting information.

Manipulation Checks
At the end of the survey, the participants in the two advantaged group conditions reflected on the anticipated presence of advantaged group members.More specifically, the participants expressed their agreement (−3 = Strongly disagree to +3 = Strongly agree) with two items asking whether White Americans/men would be there to support and stay in the background of the protest and two items asking whether the White Americans/men would be taking over the protest and be at the forefront of the protest.The manipulation was successful.The presence of advantaged group members was perceived as more supportive in the supportive-role condition in contrast to the leading-role condition.In contrast, their presence was perceived as taking over the movement in the leadershiprole condition in contrast to the supportive-role condition.More details about manipulation check analyses can be found in the Appendix S4 in the online supporting information.

Attention Checks
Lastly, the participants responded to two attention checks.The first check asked about the composition of the protesters like in Studies 1A-1C.The second check asked about who was supposed to lead the protest: (1) only disadvantaged group activists, (2) several advantaged group activists and 1 disadvantaged group activist, (3) I don't remember.Including Advantaged Groups in Collective Action

Main Analyses
We ran a Context (Racism vs. Sexism) by Audience (Disadvantaged group vs. Advantaged group) by Protest Manipulation (Control vs. Advantaged Group-Supportive role vs. Advantaged Group-Leadership role) univariate analysis of variance on all variables.Descriptive and inferential statistics are reported in Appendix S4 in the online supporting information.For reasons of brevity, we report all the significant effects, but we only describe in the detail the findings related to the manipulation and the hypotheses.
Next, we analyzed perceptions of norms regarding the engagement of advantaged group members and obtained a significant main effect of Protest Manipulation F(2,699) = 8.29, p < .001,η p 2 = .02.Replicating the findings from Study 1C, the observers perceived the norm to be stronger in the two advantaged group conditions in contrast to the control condition: M supportiverole = 4.76, SE = .09vs. M leadershiprole = 4.68, SE = .09vs. M control = 4.28, SE = .09;both contrasts were significant, p = .001,95% CI [.18, .79]and p = .005,95% CI [.10, .70].There was also a significant two-way Protest Manipulation x Audience interaction F(2,699) = 4.33, p = .014,η p 2 = .01,which suggested that the manipulation had a stronger impact on disadvantaged group observers than on advantaged group observers.More details can be found in Appendix S4 in the online supporting information.No other effects were significant.
Next, we ran a parallel mediation analysis with norm perceptions and peacefulness as mediators and politicized identification as the outcome using PROCESS macro model 4 (see Figures 4 and 5).We included two contrasts between the two advantaged group conditions and control condition as predictors, and we controlled for the effects of Context and Audience in the analyses.Bootstrapping analyses with 10,000 samples (Preacher, Rucker, & Hayes, 2007)    Importantly, contrast between supportive versus control condition was mediated via perceived norms regarding the advantaged group' engagement, B = .19,SE = .05,95%CI [.09, .29],perceived norms for disadvantaged group, B = .13,SE = .04,95%CI [.05, .22],and peaceful protest expectations, B = .08,SE = .04,95%CI [.01, .16].Contrast between leadership versus control condition was mediated via perceived norms for advantaged group, B = .15,SE = .05,95%CI [.06, .27],and peaceful protest expectations, B = .10,SE = .04,95%CI [.04, .18].The bootstrapped confidence intervals did not contain zero indicating that all indirect effects were significant.This means that the protest including advantaged group members increased politicized identification for disadvantaged and advantaged group observers by strengthening the norm perceptions and the expectations that the protest will be peaceful.Interestingly, when the observers believed that advantaged group member would be there in a supportive role, this additionally strengthened the norm that disadvantaged group members are supportive of the protest, which increased politicized identification.In contrast, this norm was not activated and did not act as a mediator when the observers were informed that advantaged group activists would be leading the protest.
We also examined the effects for disadvantaged and advantaged group separately, given that Hypotheses 2A-2B concerned the perceptions of disadvantaged group observers (see Appendix S4 in the online supporting information).In line with Hypothesis 2A, we found that advantaged group members exert a positive influence on disadvantaged group observers when they act as supporters.When advantaged group members take a leadership role their impact is diminished, but they also do not seem to have a demobilizing effect on disadvantaged group observers.All the findings reported above were replicated providing further support for the assumption that disadvantaged group observers pay attention to the role advantaged group members have in the movement.The effects on advantaged group observers were in the same direction, but much smaller and often not significant supporting the idea that advantaged group observers care less about the role allies play in a movement.

General Discussion
Theory and practice of collective action agree that in order to mobilize public support for their goals, social movement need to create a shared politicized identity with their audience (Simon & Klandermans, 2001).This project examined whether the inclusion of advantaged group members in collective action can act as a catalyst for this process and whether the process may differ for the observers from the disadvantaged and advantaged groups.First, Studies 1A-1C found that including equal numbers of disadvantaged and advantaged group members in collective action (as opposed to collective action without any advantaged group members) increased politicized identification among advantaged group observers, as predicted by Hypothesis 1.Second, Study 2A-2B revealed that when advantaged group members take a supportive role in the protest, disadvantaged group observers were also more likely to increase their politicized identification in line with Hypothesis 2A.We did not find support for Hypothesis 2B which suggested that the presence of advantaged group members has negative effects on politicized identification among disadvantaged group observers, even if they take a leadership role at a protest.Third, Studies 1C-2A-2B revealed that the positive effects on politicized identification were driven by perceived norms and expectations that the protest will be peaceful.

Theoretical Implications
The present work contributes to the growing literature on solidarity and allyship.Our studies suggest that the inclusion of advantaged group members can be beneficial for social movements, because it facilitates the identification with and the internalization of the movement's cause (Subašić et al., 2008).Importantly, we build on previous work by showing that this process differs depending on the audience movements target.For advantaged group observers, a large presence of advantaged group allies increased politicized identification by strengthening the solidarity norms in line with the research on normative influence (Louis, 2009).Moreover, a protest including advantaged group members was perceived as more likely to remain peaceful, which fits with work on public perceptions of allies who confront discrimination (Czopp & Monteith, 2003).In addition to expecting the protest to be peaceful, the observers may have inferred that the protesters' demands are less likely to be radical and extreme.We expect that this should be the case for advantaged group observers who want to protect their group's privileged status and perhaps believe that movements including allies will refrain from seeking radical structural changes.
Even though the increase in politicized identification seems as a promising step in mobilizing advantaged groups, this does not necessarily mean that more advantaged group members will take to the streets and show their support.Increasing politicized identification is only the first step in a long process of mobilization for social change (Simon & Klandermans, 2001).However, social movements that are able to secure a larger pool of potential followers have better chances of getting some of them to publicly show their support.Moreover, it is not clear whether the observers believe that the movements including advantaged group allies are more likely to be effective in securing societal changes as suggested by theoretical work (Subašić et al., 2008).Future research could examine whether the general public believes that including advantaged group allies puts more pressure on power holders, as well as whether power holders are more likely to compromise and support policies proposed by a movement including allies than by a movement that does not.However, we warn against the tendency to see advantaged group allies as saviors, because the inclusion of allies might result in less support for the movement among disadvantaged groups (Iyer & Achia, 2020) or may force movements into compromises that fall short of true social change (Droogendyk et al., 2016).
The findings among disadvantaged group observers suggest that they care about the role advantaged group allies play in the protest, which fits with work by Iyer and Achia (2020) in the organizational context.Our findings complement this work by showing that the presence of allies at protests has positive effects on disadvantaged group's identification with the movement in contrast to when allies are not included.However, this positive effect disappears when allies take a leadership role, although we did not find evidence that disadvantaged group observers disidentify from the movements where advantaged group members take the lead.In contrast to other work, we only mentioned a single protest occasion in our studies, and we expect that disadvantaged group observers would react more negatively to advantaged group activists occupying long-term leadership positions within a social movement.
Lastly, our studies speak to the importance of group norms in facilitating social change.Crosscultural research on prejudice (Visintin, Green, & Sarrasin, 2018) and volunteering (Roblain, Hanioti, Paulis, Van Haute, & Green, 2020) finds that broader normative climate and peer norms shape individuals' motivations and attitudes towards disadvantaged groups.Building on this work, we show that solidarity and participation norms can be strengthened by including advantaged group allies.An intriguing question for future research is to examine whether advantaged group allies signaling descriptive versus injunctive norms secure a stronger bond with observers.Although we did not differentiate between the two types of norms, research suggests that injunctive norms should have a stronger impact on politicization and mobilization for collective action (Roblain et al., 2020).

Limitations
We note that the obtained effects across all studies are small, and we encountered a relatively substantive number of participants who did not fill out the survey seriously and/or failed the attention checks.There is concern over the quality of the data obtained via online platforms like Amazon's

Figure 1 .
Figure 1.Conceptual model for advantaged and disadvantaged group observers.

Figure 2 .
Figure 2. Black Americans' and White Americans' politicized identification in Study 1A.Error bars represent standard errors.

Figure 3 .
Figure 3. Women's and men' politicized identification in Study 1B.Error bars represent standard errors.