Conditional Parochial Vicarious Ostracism: Gender Collective Narcissism Predicts Distress at the Exclusion of the Gender Ingroup in Women and Men

Three experiments (two pre-registered) tested whether gender collective narcissism (i.e., a belief that one’s gender ingroup’s exceptionality is not sufficiently recognized by others) predicts parochial vicarious ostracism (i.e., feeling ostracized and distressed while recognizing the gender ingroup’s exclusion, but not when recognizing the exclusion of the gender outgroup). In all studies (overall N = 1480), gender collective narcissism was positively associated with distress among women who witnessed the exclusion of women, but not among men who witnessed the exclusion of women. In Study 3, gender collective narcissism was positively associated with distress among men who witnessed the exclusion of men, but not among women who witnessed the exclusion of men. These findings help explain why men do not universally feel distressed by the discrimination of women and why some women may mobilize to challenge gender discrimination.

The recent over-ruling of the Roe vs. Wade decision, which has guaranteed women's reproductive rights in the U.S. since 1973, illustrates the real threat to gender equality under the current global wave of ultraconservative populism (Graff & Korolczuk, 2021;Juhasz, 2018). Gender backlash and discrimination is costly for women (Mills et al., 2020;Milner et al., 2020), who may experience vicarious distress from gender discrimination even when its cues are subtle and indirect (McCarty et al., 2022). The role of women's gender identification in experiencing distress at the exclusion of other women is still unclear. Research has shown that gender identification may ameliorate (Bolling et al., 2012) or aggravate (McCoy & Major, 2003) women's distress from gender discrimination depending on the aspect of gender identification being assessed (Schmitt et al., 2014). Similarly, the role of gender identification among men witnessing the exclusion of women is understudied. Although men allies are important in challenging gender inequality (Moser & Branscombe, 2022;Warren et al., 2021), their solidarity with women may depend on how men identify with their gender ingroup (DiMuccio & Knowles, 2020;Golec de Zavala & Bierwiaczonek, 2021;Górska et al., 2020;Kosakowska-Berezecka et al., 2020).
More generally, the allyship of advantaged groups (e.g., men) with disadvantaged groups (e.g., women) is motivated by members of advantaged groups feeling distressed and angered by the discrimination of disadvantaged groups (Pereira et al., 2022;Subašić et al., 2008). In both groups, emotional reactions to discrimination are shaped by normative prescriptions embedded in the way group members construe their social identity (Thomas et al., 2009;van Zomeren et al., 2018). In particular, collective narcissism, which expresses one of the ways people construe their social identity, is a robust predictor of endorsing discrimination (for reviews see Golec de Zavala, 2011;Golec de Zavala et al., 2019;Golec de Zavala & Lantos, 2020), and gender discrimination, specifically (Golec de Zavala & Bierwiaczonek, 2021;, 2022a. Collective narcissism is a belief about an ingroup that can be applied to any ingroup people identify with (Golec de Zavala et al., 2009; for review and discussion, Golec de Zavala, 2011). Gender collective narcissism refers to the belief that the gender ingroup's exaggerated importance and exceptionality are not sufficiently recognized by others (Golec de Zavala & Bierwiaczonek, 2021;Golec de Zavala et al., 2009). The present research extends prior work on the association between gender collective narcissism and gender discrimination (Golec de Zavala & Bierwiaczonek, 2021, Golec de Zavala & Keenan, 2022b, for review see Golec de Zavala & Keenan, 2022a) by examining the proposition that gender collective narcissism enhances the disparity in men's and women's distress in response to gender-based exclusion. That is, it facilitates distress at the exclusion of the gender ingroup but constrains distress at the exclusion of the gender outgroup. This proposition is supported by findings that gender collective narcissism predicts attitudes sustaining gender inequality among men, but attitudes challenging gender inequality among women (Golec de Zavala & Keenan, 2022b).
Further, the current paper focuses on the exclusion of the gender ingroup as a specific aspect of gender discrimination. Integrating the literature on vicarious ostracism (i.e., feeling ostracized and distressed while recognizing that someone else is ignored or excluded without personally experiencing first-hand exclusion; Wesselmann et al., 2013), and parochial empathy (i.e., greater empathy for distress and suffering of members of the ingroup vs. outgroup; Bruneau et al., 2017;Cikara et al., 2011), I predict that men and women may experience distress in the face of exclusion of their own gender ingroup, rather than the gender outgroup. In other words, the intergroup context may limit vicarious ostracism to ingroup members only. I expect this pattern would be more likely at higher levels of gender collective narcissism.

Parochial Vicarious Ostracism and Gender Group Exclusion
As members of a disadvantaged social group, women experience distress from gender discrimination (Bilodeau et al., 2020;Matheson et al., 2019). Women feel distressed when they witness the exclusion of other women even without being excluded themselves (McCarty et al., 2022;Schmitt et al., 2002). Women's distress when personally excluded intensifies when women attribute personal exclusion to their group membership (Schaafsma & Williams, 2012;Schmitt et al., 2003;Wirth & Williams, 2009). Social exclusion-being ignored or rejected by others against one's will-is a universally painful experience (Hartgerink et al., 2015), and witnessing someone else being excluded is often as painful as being personally excluded (Wesselmann et al., 2013). Why then do some men behave as if they do not experience distress at the exclusion of women?
One explanation is that people tend to empathize more with the distress of members of the ingroup than the outgroup (Cikara et al., 2011), and even rejoice in the distress of the outgroup (Cikara & Fiske, 2012;Cikara et al., 2014). Indeed, people experience vicarious ostracism more when they witness the exclusion of those who are similar to them (Veldhuis et al., 2014), and with whom they are able to empathize (Beeney et al., 2011;Wesselmann et al., 2013). Thus, in intergroup contexts, vicarious ostracism may be parochial. This proposition has never been studied in the context of gender groups. I propose that women may feel distressed when witnessing the exclusion of women because being a woman is a part of their identity and thus exclusion of women is psychologically consequential for them even when a woman witnessing the exclusion of other women is not personally excluded. Men may be less likely than women to be distressed when witnessing the exclusion of women because they construe women as the outgroup. These parochial responses to exclusion are more likely at higher levels of gender collective narcissism, a variable that predicts perceiving the relations between gender groups as a zero-sum conflict (Golec de Zavala & Keenan, 2022a).
The parochial responses to the ingroup's exclusion are likely to be symmetrical among men and women. Although men experience discrimination less often than women (Manzi, 2019), they are likely to feel distress at the exclusion of other men and also experience it as social identity threat. Members of advantaged groups experience distress even when their ingroup is momentarily excluded in an abstract game of tossing a virtual ball Hase et al., 2021). Moreover, while members of disadvantaged groups (like women) are distressed by reminders of their ingroup's exclusion (McCarty et al., 2022), members of advantaged groups (like men) feel distressed at the prospect of their ingroup's exclusion (Scheepers, 2009). This prospect is especially salient when members of advantaged groups construe intergroup equality as exclusion of their ingroup from power and privilege (Inglehart & Norris, 2016;Jetten, 2019;Mutz, 2018), and exacerbated by higher collective narcissism . Indeed, the current backlash against the pursuit of gender equality is spearheaded by men fearing exclusion from group-based privilege (Flood et al., 2020;Graff & Korolczuk, 2021). Those men also tend to endorse gender and national collective narcissism (Golec de Zavala & Bierwiaczonek, 2021;Górska et al., 2020).

Gender Collective Narcissism and Parochial Vicarious Ostracism
Gender collective narcissism may predict parochial distress at the exclusion of the gender ingroup among men and women alike. Gender collective narcissism among men and women is associated with the perception of the gender ingroup's relative deprivation and a zero-sum conflict between men and women (Golec de Zavala & Keenan, 2022b). Among men, gender collective narcissism is associated with the legitimization of the gender hierarchy (Golec de Zavala & Keenan, 2022b), and the belief that manhood is precarious and non-traditional women threaten men's social status (Golec de Zavala & Bierwiaczonek, 2021;Górska et al., 2020). Thus, men who endorse gender collective narcissism may be particularly unlikely to feel distressed at the exclusion of women but likely to feel distressed when witnessing the exclusion of other men. As the predictions of collective narcissism can generalize beyond particular group memberships and intergroup contexts (Golec de Zavala et al., 2009; for review see Golec de Zavala, 2011;Golec de Zavala et al., 2019), gender collective narcissism among women is also likely to predict greater distress at the exclusion of women vs. men. This prediction is supported by findings that gender collective narcissism predicts support for egalitarian ideology and collective action towards gender equality (Golec de Zavala & Keenan, 2022b).
The above predictions regarding the role of gender collective narcissism are at odds with the Rejection Identification Model (Branscombe et al., 1999a, b), which suggests that ingroup identification should play a palliative role in the face of aversive experiences such as social exclusion. However, empirical results regarding the Rejection Identification Model are inconsistent (Schmitt et al., 2014). Women's gender identification is linked to greater physiological and psychological distress after exposure to gender discrimination (Eliezer et al., 2010;Fischer & Holz, 2007;McCoy & Major, 2003). Such findings are in apparent conflict with results suggesting that women's private collective self-esteem (i.e., positive evaluation of their gender ingroup) buffers the negative effects of gender discrimination (Corning, 2002), and affirmation of gender ingroup values may protect women's self-esteem in the face of blatant sexism (Spencer-Rodgers et al., 2016). Differentiating gender collective narcissism from non-narcissistic collective self-esteem or ingroup satisfaction may help reconcile those seemingly contradictory findings.
While positive ingroup identification provides psychological resources (e.g., clear self-definition, high self-esteem, a sense of meaning and direction, a sense of belonging, and social connectedness) that support individual wellbeing (Cruwys et al., 2014;Jetten et al., 2017), the positive and protective role of sharing a social identity depends on its normative content, the ingroup's status and circumstances. For example, group norms may require group members to engage in actions that undermine their wellbeing (e.g., violence in gangs). Similarly, collective narcissism emphasizes under-appreciation of the ingroup by others. When it becomes a prevalent way of defining the ingroup's identity, it is likely to undermine group members' wellbeing. Indeed, collective narcissism is associated with chronically low life satisfaction, predominantly negative mood and negative emotionality (Golec de Zavala, 2019;Golec de Zavala et al., 2022), and, unlike non-narcissistic ingroup satisfaction, it does not predict well-being in advantaged or disadvantaged groups (Bagci et al., 2021). Thus, to understand the role of ingroup identification in the context of intergroup exclusion, it is important to examine not only whether group members identify with the excluded ingroup, but also how they identify with it, as ingroup identification is a multifaceted phenomenon (e.g., Leach et al., 2008).
As an individual difference variable, collective narcissism can characterise group members with a relative degree of stability in reference to the same social identity (as indicated by longitudinal studies; Federico et al., 2021;, and across intersecting social identities (as indicated by studies comparing collective narcissism with reference to different social identities; Golec de Zavala & Keenan, 2022b;Mole et al., 2021). However, like group membership salience and ingroup identification (Mullen et al., 1992;Thomas et al., 2020), collective narcissism is higher among members of disadvantaged groups in comparison to members of advantaged groups (Bagci et al., 2021;Golec de Zavala & Keenan, 2022a;Golec de Zavala et al., 2009).
In advantaged groups enjoying high status and privilege, the claims regarding the lack of sufficient recognition of the ingroup's importance are narcissistic, exaggerated, and delusional. However, in disadvantaged groups, the same claims are not narcissistic, exaggerated, or delusional. Disadvantaged groups do lack recognition in comparison to advantaged groups and the claim to be recognized as equal is not itself narcissistic. Yet, members of disadvantaged groups can endorse collective narcissism with reference to their ingroup. Even in disadvantaged groups collective narcissism expresses the need to be recognized as better than others, not as equal to others. Indeed, across various group memberships and levels of social status (i.e., in advantaged and disadvantaged groups), collective narcissism predicts intergroup antagonism and hostility towards outgroups (Golec de Zavala & Lantos, 2020;Golec de Zavala et al., 2016 as well as the acceptance of political violence or belligerent, non-normative collective action to advance the ingroup interests (Golec de Zavala & Keenan, 2022b;Jasko et al., 2020;. Given that it facilitates the perception and escalation of intergroup conflict (Golec de Zavala et al., 2009, gender collective narcissism is likely to be associated with greater disparity in men and women's reactions to the exclusion of their respective gender ingroup vs. outgroup. In other words, while gender collective narcissism among women is likely to predict greater distress at the exclusion of women but not the exclusion of men, gender collective narcissism among men is likely to predict distress at the exclusion of men but not the exclusion of women. Gender collective narcissism may even predict rejoicing at the distress of the exclusion of the outgroup, as previous findings have linked collective narcissism to opportunistic intergroup aggression in the form of Schadenfreude (Golec de Zavala et al., 2016).

Overview
In Study 1 and 2 (Study 2 was pre-registered), I tested whether gender collective narcissism is positively associated with distress when women witness the exclusion of other women (H1), and negatively associated with distress when men witness the exclusion of women (H2). In Study 3 (preregistered), I tested whether gender collective narcissism is also positively associated with distress when men witness the exclusion of other men (H3), and negatively associated with distress when women witness the exclusion of men (H4). I specified all hypotheses comparing the association of gender collective narcissism and distress of exclusion of the gender ingroup to the association of gender ingroup identification and distress of exclusion of the gender ingroup.
I tested all hypotheses in Poland, where women's reproductive rights have been systematically limited since the 1990s and overtly attacked since the ultraconservative populist government came to power in 2015 (Graff & Korolczuk, 2021). The Women's Strike involved women's nationwide protests in 2020 and has been met with violent responses from the state (Human Rights Watch, 2021). In Study 1, I designed the experimental manipulation around those events. In Study 2 and 3, I used Cyberball-a controlled, context-free, and internally valid paradigm-to manipulate interpersonal (Hartgerink et al., 2015) and intergroup exclusion Hase et al., 2021).
All studies capitalize on analyses supporting the measurement invariance of the Collective Narcissism Scale to assess the same construct in women and men, and therefore predictions of gender collective narcissism can be compared between those groups (Golec de Zavala & Keenan, 2022b). In all studies, gender ingroup identification was also tested as an alternative continuous predictor to determine whether the expected associations are specific to gender collective narcissism. Opposite expectations regarding gender ingroup identification can be derived from the Rejection Identification Model (Branscombe et al., 1999a, b), which predicts that the association between gender identification and distress at the exclusion of the gender ingroup should be negative.
In Study 2, H1-H4 were specified against individual narcissism, which I examined as an alternative predictor of parochial vicarious ostracism. Previous studies show that individual narcissism enhances distress among participants who experienced exclusion first-hand (Blinkhorn et al., 2021;Twenge & Campbell, 2003). However, evidence also indicates that collective, not individual, narcissism predicts intergroup attitudes and behaviors (Golec de Zavala et al., 2009, 2022Hase et al., 2021). These analyses were not pre-registered, but they help to illustrate the unique role of collective narcissism in predicting reactions specific to intergroup context (Golec de Zavala et al., 2009, 2022. Those analyses yielded different patterns of results than analyses with gender collective narcissism. Individual narcissism was not associated with parochial vicarious ostracism. For the sake of brevity, they are not presented in detail in the main text (codes for those analyses are available at https:// osf. io/ 83d45/).
In Study 3, to test whether the results generalize across different methods of assessment of distress (i.e., the negative affect sub scale of the Positive and Negative Affect Scale, Watson et al., 1988) was used to provide exact and conceptual replication of the results from Study 1 and 2. In all studies, vicarious ostracism and parochial vicarious ostracism were tested first, by comparing distress at the exclusion of the gender ingroup to distress at the exclusion of the outgroup, among men and women. A main effect suggesting an increase in distress after witnessing the exclusion of any gender group member would be indicative of vicarious ostracism. The moderation of this effect by gender (e.g., men feel less distress than women when witnessing exclusion of women) would be indicative of parochial vicarious ostracism.
Sample size estimations were carried out using G*Power (Faul et al., 2009). To test and specify H1 and H2, I estimated a sample size required for a multiple regression analysis with 7 predictors. I entered f 2 = 0.05 for the increase in the percentage of explained variance based on previous research on collective narcissism and distress at the ingroup's exclusion (Hase et al., 2021). To test H3 and H4, I performed the same analysis for 11 predictors. For 80% power at α = .05, the required sample to test and specify H1 and H2 was 295, for H3 it was 346.

Study 1
We conducted Study 1 in the context of nationwide protests that took place after the near total ban on abortion was introduced in Poland in October 2020. The experimental manipulation was designed based on a real-life event in the Polish parliament. On October, 22, 2020, while female MPs (led by Joanna Scheuring-Wielgus) spoke in support of women's protests to the parliament, the microphone was switched off by the male Chairman of the Parliament (Ryszard Terlecki). This event presented an opportunity for our ecologically valid manipulation of vicarious effects of the exclusion of women. Study 1 tested H1 and H2 using distress in response to the exclusion of a female MP as the outcome variable. Support for the women's strike was treated as a covariate likely to account for variance in distress. Following the recommendations of Simmons et al. (2011), I first performed the analyses without the covariate. The omnibus results for the tests of H1 and H2 were not significant (p = .17) for the hypothesized three-way interaction, whereas the expected association between women's gender collective narcissism and distress at the exclusion of the gender ingroup was significant (codes available at https:// osf. io/ 83d45/). However, the expected three-way interaction became significant when the covariate was added to the analysis. In other words, the three-way interaction explained additional variance in distress net of support for the women's strike. For the sake of brevity, I present only the significant results.

Participants
The participants were a convenience sample of 407 Polish adults (290 women and 117 men, age ranged from 18 to 71 years, M = 25.49, SD = 8.28). The sample consisted predominantly of students who participated and further recruited up to 10 additional participants in exchange for research participation credits as a part of a research method assignment. Initially, 1,461 participants passed the technical test (to watch the videos on their computers) and were randomly allocated to the research conditions by the Qualtrics platform (qualtrics.com), which supported the data collection. Of the registered participants, 695 (48%) provided consent and survey responses (467 (67%) women and 216 (31%) men, 5 (0.7%) non-binary, 7 (1.3%) did not answer; age ranged from 18 to 71 years, M = 25.93, SD = 8.59). Prior to analyses, data were excluded from 84 (12%) participants: 72 participants who provided incorrect answers to an attention check question (i.e., Please choose answer 'Agree') and 12 participants who did not indicate their gender or indicated non-binary gender identification. Data from a further 204 (29%) participants with repeated IP addresses were excluded as it was impossible to determine whether responses came from different participants or the same participant responding to the survey several times. The results showed the same pattern without these exclusions; however, I chose to exclude these cases from the analyses to err on the side of caution.

Materials
Unless otherwise indicated, items were scored on a 6-point scale ranging from 1 (completely disagree) to 6 (completely agree). Higher scores indicate higher levels of the assessed variables.
Distress was assessed by six items validated in previous research on ostracism (Gonsalkorale & Williams, 2007) and previous research in Poland (Hase et al., 2021). We asked participants to indicate to what extent they felt each of following states while observing the video: "good" (reversed), "happy" (reversed), "relaxed" (reversed), "resentful", "upset", and "threatened". The responses to those items were averaged after the responses to the first three items were reversed. Higher scores reflect more distress, α = .86.
Gender collective narcissism was assessed using the five-item Collective Narcissism Scale, with reference to the gender ingroup (Golec de Zavala et al., 2009). Participants responded to the following items that instructed them to think about their respective gender group when responding: "If my group had more to say in the world, the world would be a much better place"; "My group deserves special treatment"; "I get angry when others criticize my group"; "Not many people understand the true importance of my group"; and "I will not be satisfied until my group gets the recognition it deserves." The items were averaged so that higher scores reflect higher gender collective narcissism, α = .86.
Gender ingroup identification was assessed with the 16-item In-Group Identification Scale (Leach et al., 2008), used in previous studies in Poland  to measure the following components of gender identification: solidarity with one's gender group (3 items, e.g., "I feel solidarity with my gender group"), satisfaction with being a member of the gender group (4 items, e.g., "I think that my gender group has a lot to be proud of"), the centrality of one's gender group (3 items, e.g., "I often think about the fact that I am a member of my gender group"), individual self-stereotyping (3 items, e.g., "I have a lot in common with the average member of my gender group"), and in-group homogeneity (3 items, e.g., "Members of my gender group have a lot in common with each other"). All items were completed using the 6-point scale described above. As the subscales were highly intercorrelated, for the sake of clarity and brevity, responses were averaged across items so that higher scores reflect higher gender ingroup identification, α = .94. Support for the All-Poland Women's Strike was assessed with the following question: "Do you support the All-Poland Women's Strike?" The responses ranged from 1 (definitely no) to 6 (definitely yes). Higher scores indicate more support for the strike.
Manipulation check was assessed by four statements: "While watching the video, I had the impression that my group was excluded," "While watching the video, I had the feeling that my group was being ignored," "I felt excluded while watching the video," and "I felt ignored while watching the video." A maximum likelihood exploratory factor analysis indicated a one factor solution with items loadings between .85-.92. The 4-item scale had high reliability, α = .93, M = 3.44, SD = 1.41.

Procedure
The study followed a 2 (exclusion: control vs. exclusion of women) × 2 (gender: men vs. women) design. Men and women were randomly assigned to the exclusion conditions. Participants gave their informed consent to take part in the study that ostensibly examined the link between action and ability of mental visualization. They were asked to watch a short video and observe their thoughts and feelings. Before the experimental manipulation was introduced, participants were asked to respond to demographic questions, including support for the women's strike, and a measure of national collective narcissism and ingroup satisfaction (those measures were collected for a different project and not analyzed here).
In the exclusion of women condition (n = 140), participants watched a video showing the parliamentary session, during which a female MP (Joanna Scheuring-Wielgus) spoke on behalf of the All-Poland Women's Strike in support of women's reproductive rights. In this video, male Chairman of the Parliament (Ryszard Terlecki) switched off the microphone while the female MP was still speaking. In the control condition (n = 267), participants watched a video showing a parliamentary session with the same female MP speaking on the same topic, but her microphone was not switched off. After the manipulation, participants completed questions to check the validity of the manipulation and distress was measured. The order of the distress items was randomized for each participant. Next, the measures of gender collective narcissism and gender identification were presented in a random order, with the order of items was also randomized for each participant. At the end of the study, participants were asked to guess the purpose of the study (nobody guessed), debriefed, and thanked.

Study 1 Results and Discussion
Means, standard deviations, and zero-order correlations among all variables in Study 1 are presented in Table 1. All variables were positively correlated with each other.
In order to establish the effectiveness of the experimental manipulation of the exclusion of women, a univariate 2 × 2 ANOVA was performed to determine the effect of witnessing the ingroup's exclusion on perceived personal and ingroup exclusion. As the results in Table 2 indicate, the significant main effects of the experimental manipulation and gender were qualified by a significant interaction. Men and women felt more excluded (personally and as a group) in the exclusion vs. control condition, but women felt more excluded than men in the exclusion condition. This pattern of results suggests that the sense of exclusion in the intergroup context was evoked successfully.
In Study 1 and 2, gender collective narcissism and gender identification were assessed after the experimental manipulation to test the alternative prediction, derived from the  (Branscombe et al., 1999a, b), that witnessing gender exclusion should increase gender identification and, in turn, protect group members from the adverse effects of exclusion. The results of 2 × 2 ANOVAS indicated no effect of the experimental manipulation (alone or in interaction with gender) on gender collective narcissism (ps > .34) or gender ingroup identification (ps > .10; codes available at https:// osf. io/ 83d45/). Thus, gender collective narcissism and gender identification were analyzed as continuous moderators. This strategy is in line with the conceptualization of gender collective narcissism as a relatively stable individual difference variable that does not fluctuate in response to momentary characteristics of the situation (Golec de Zavala et al., 2009. The data were analyzed for a pattern indicative of vicarious ostracism; that is, greater distress in the exclusion condition in comparison to the control condition. This was done to test the alternative expectation, derived from vicarious ostracism theory (Wesselmann et al., 2013), which argues that social exclusion is a universally painful experience, and we feel ostracized and distressed when witnessing others being socially excluded. Accordingly, pain of exclusion should be experienced vicariously even when one witnesses exclusion of someone else without being personally excluded and regardless of the excluded person's social identity.
Next, the data were analyzed for a pattern indicative of parochial vicarious condition; that is, that vicarious ostracism may be a function of gender group membership. In line with this proposition women's distress would be expected at the exclusion of other women (the gender ingroup) but not at the exclusion of men (the gender outgroup). The expectation that women may feel excluded only when witnessing the exclusion of women (but not when witnessing the exclusion of men) was derived from the parochial empathy research suggesting that people resonate with the suffering of members of their ingroup but not with the suffering of members of the outgroup (e.g., Bruneau et al., 2017); and qualifies the predictions derived from the vicarious ostracism literature.
The results of a 2 × 2 ANOVA testing for vicarious ostracism and parochial vicarious ostracism indicated the presence of neither form of ostracism (see Fig. 1). The main effect of exclusion on distress was not significant, p = .28. The interaction of gender and exclusion was not significant, p = .73. Distress varied only as a function of gender. Distress was higher among women in comparison to men across the exclusion conditions, F(1, 403) = 33.51, p < .001, ω 2 = .07. The results were the same when support for the women's strike was entered as a covariate. Thus, there was no evidence of vicarious ostracism (distress as a function of witnessing the exclusion of women among men and women alike) or parochial vicarious ostracism (i.e., distress as a function of witnessing the exclusion of women only among women) in Study 1.

Moderated Moderation to Test H1 and H2
H1 and H2 were tested with a moderated moderation analysis using a hierarchical multiple regression. A three-way interaction was hypothesized between gender collective narcissism (centered), gender group (0 = women, 1 = men), and exclusion (0 = control, 1 = exclusion of women) in predicting distress (Hayes, 2022, Model 3). Support for the women's strike was entered as a covariate (centered). We requested 10,000 bootstraps. In all studies, analyses that are robust to the violation of the assumption of homoscedasticity of the residuals were performed (HC3, as recommended by Hayes & Cai, 2007, codes available at https:// osf. io/ 83d45/). The significant model with three main effects, three two-way interactions, and one  Table 3). While gender and exclusion did not significantly affect distress, gender collective narcissism was positively associated with distress. The three-way interaction between gender, exclusion, and gender collective narcissism was significant, p = .039 (see Fig. 2). Table 4 presents the simple slopes for men and women in each condition. Consistent with H1, gender collective narcissism was positively associated with distress among women who witnessed the exclusion of women, but not in the control condition. With respect to H2, gender collective narcissism was negatively associated with distress when men witnessed the exclusion of women and positively associated with distress in the control condition. However, none of the associations reached the conventional statistical significance level among men.
To specify H1 and H2, the same analyses were performed entering gender identification instead of gender collective narcissism as a predictor of distress at the exclusion of the gender ingroups. Drawing on the Rejection Identification Model (Branscombe et al., 1999a, b), distress should be lower among women (vs. men) who score higher on gender identification. The analyses revealed a positive association between support for the women's strike and distress, b = .24; SE = .03; t(398) = 6.99; p < .001. The role of gender identification was also significant in predicting distress across the exclusion conditions, b = .12; SE = .05; t(398) = 2.32; p = .02. However, there was no significant moderation of the association between gender identification and distress by gender or exclusion, b = -.17; SE = .23; t(398) = 0.75; p = .45. Thus, while gender collective narcissism predicts distress when witnessing the exclusion of the gender ingroup specifically, gender identification predicted greater distress in general.
While the results of Study 1 show, as expected, that gender collective narcissism predicts women's (but not men's) distress at the exclusion of women, the study had several shortcomings. Although the experimental manipulation was ecologically valid, it might have been confounded with participants' political outlook: the female MP represented the leftist political party, whereas the Chairmen represented the rightist party. In addition, the number of participants allocated to Study 1 and Study 2 was unequal. Thus, Study 2 was conducted to better disentangle political orientation from the exclusion of women within the experimental manipulation and tested H1 and H2 in a more neutral setting and with better control over balanced allocation to study conditions.

Study 2
Study 2 used an internally valid manipulation of gender exclusion, the Cyberball paradigm (Williams & Jarvis, 2006), to test H1 and H2. The hypotheses, design and analytic strategy were pre-registered (https:// aspre dicted. org/ 3fx2a. pdf). For consistency with Study 1, we performed analyses with support for the women's strike as a covariate. Analyses with this covariate were not preregistered and they yielded the same pattern of results as without the covariate (available at https:// osf. io/ 83d45/). Study 2 also tested H1 and H2 with gender identification and individual narcissism as alternative moderators (those analyses were not pre-registered).

Participants
The participants were a convenience sample of 385 Polish adults (242 women and 142 men; age ranged from 18 to 67 years, M = 29.47, SD = 9.02). Data were collected in the same manner as in Study 1. The samples in Study 1 and 2 are independent. Studies were conducted in different cohorts and participants in Study 1 were prevented from participation in Study 2. No IP addresses overlap between the two studies. Only participants who were not familiar with the Cyberball paradigm could participate in the study.
Initially, 845 participants registered to participate in the study and passed the technical test to assure they were able to watch the Cyberball game. They were randomly allocated to the research conditions by the Qualtrics platform (qualtrics.com), which hosted the study. Out of all registered participants, 47 could not continue the study because they did not indicate whether they were familiar with the Cyberball paradigm. Of the 798 (94%) eligible participants, 517(65%) provided data and consent (305 (59%) women and 196 (38%) men, 13 (3%) non-binary, 3 (.01%) did not answer; age ranged from 18 to 71 years, M = 30.05, SD = 9.66). Prior to analyses, data from 36 (7%) participants were excluded based on the pre-registered criteria: 21 participants who provided wrong answers to one attention check question (e.g., 'Please choose answer 'Agree''), and 15 participants who indicated a non-binary gender identification. Additionally, data from 96 (18%) participants with repeated IP addresses were excluded as in Study 1. This exclusion criterion was not pre-registered as this problem was not foreseen.

Materials
Unless otherwise indicated, items were scored on a 6-point scale ranging from 1 (completely disagree) to 6 (completely agree). Higher scores indicate higher levels of the assessed variables.
Distress (α = .89), gender collective narcissism (α = .85) and gender ingroup identification (α = .93) and the one item measure of support for All Poland Women's Strike were assessed as in Study 1.
Vulnerable narcissism was assessed by the short version of the Narcissistic Vulnerability Scale (Crowe et al., 2018) used in previous studies in Poland . Participants viewed a list of six adjectives ("insecure," "ignored," "envious," "unappreciated," "resentful," and "not understood") and indicated the degree to which each adjective described them using the 6-point scale described above ("Does this adjective define you?"). Higher scores indicate higher vulnerable narcissism, α = .86.
Grandiose narcissism was assessed using the Polish adaptation (Rogoza et al., 2016) of the Narcissistic Admiration and Rivalry Questionnaire (Back et al., 2013). This scale comprises 18 items that pertain to the admiration (e.g., "I am great") and rivalry ("I want my rivals to fail") aspects of grandiose narcissism and were answered using the 6-point scale described above. The positively correlated subscales of rivalry and admiration (r(383) = .45, p < .001) were collapsed for the sake of brevity. Higher scores indicate higher grandiose narcissism, α = .85.
Manipulation check questions included a standard manipulation check for the Cyberball paradigm: "What percentage of all ball throws did your group receive in the Cyberball game (choose a number between 0 and 100)?" The item was scored on a 100-point scale ranging from 0 to 100, anchored with 5-point increments (i.e., at 0, 5, 10, etc.), M = 43.53, SD = 23.91. Higher responses to this question confirm that participants perceived the exclusion accurately. The items assessing perceived personal and group exclusion as in Study 1 were also completed, α = .94, M = 3.06, SD = 1.42. Higher responses to those items indicate participants felt excluded in the exclusion condition.

Procedure
The study followed a 2 (exclusion: control vs. exclusion of women) × 2 (gender: men vs. women) design. Men and women were randomly assigned to exclusion conditions by the survey platform. Participants gave their informed consent as in Study 1. They were informed they would be assigned to either the role of a player or an observer in an online interaction between men and women. Before the experimental manipulation was introduced, participants were asked demographic questions including to what extent they support the All-Poland Women's Strike. Next, participants were led to believe that they were randomly allocated to the role of the observer in an online game. We used an adapted intergroup Cyberball paradigm (Williams & Jarvis, 2006;Williams et al., 2000) as in previous studies (Hase et al., 2021).
In the interpersonal Cyberball paradigm (Hartgerink et al., 2015;Williams et al., 2000), participants are led to believe that they are playing an online ball-tossing game with two other participants. All participants observed the Cyberball game between teams of three men and three women labeled with male vs. female names, respectively, and visually categorized by the different coloring of the avatars (different between groups, the same within the group).
Participants were asked to visualize the game as if it was happening in real-life (imagining the settings, the participants, etc.). In the exclusion condition (n = 186), participants observed women receiving 3 out of 30 ball-passes (10%) at the beginning of the game and then men exchanged the ball tosses between themselves excluding women from the game. In the control condition (n = 196), participants observed teams of men and women exchanging an equal number of ball-passes throughout the game.
After the game, participants completed manipulation check questions. Next, they indicated their level of distress, with the order of the scale items randomized as in Study 1. Finally, participants completed measures of gender collective narcissism and gender identification presented in a random order, with the order of items randomized for each participant. At the end of the study, participants were probed for guessing (nobody guessed the hypothesis or reported suspicion about the study), debriefed, and thanked.

Study 2 Results and Discussion
Means, standard deviations, and zero-order correlations among all variables in Study 2 are presented in Table 1. Gender collective narcissism, individual narcissism, and gender identification were positively correlated. Gender identification was negatively associated with vulnerable narcissism and positively associated with grandiose narcissism. Collective and individual narcissism, but not gender identification, were positively associated with distress and support for the women's strike.
To check that the manipulation of the exclusion of women was effective, I performed a univariate 2 × 2 ANOVA to determine the effect of the exclusion condition on the manipulation check questions. The significant main effects of exclusion and gender were qualified by a significant interaction (see Table 2). While men and women felt more excluded in the exclusion than in the control condition, women felt more excluded in this condition than did men. The same pattern of results was found for perceived percentage of ball-tosses in each condition.
As in Study 1, gender collective narcissism and gender ingroup identification were assessed after the experimental manipulation. The results of 2 × 2 ANOVA indicated the exclusion condition did not affect gender collective narcissism or gender identification. There were also no significant interactions between the exclusion condition and gender on gender collective narcissism or gender identification (ps > .34 and .24, respectively). Thus, gender collective narcissism and gender identification were analyzed as continuous moderators.
As depicted in Fig. 1, the results of a 2 × 2 ANOVA testing for vicarious ostracism and parochial vicarious ostracism indicated the presence of vicarious ostracism. The main effect of exclusion was significant, F(1, 381) = 97.52, p < .001, ω 2 = .20; men and women felt more distressed when witnessing the exclusion of women in the Cyberball game, relative to the control condition. In addition, as in Study 1, women felt more distress than men, F(1, 381) = 13.83, p < .001, ω 2 = .03 (Fig. 1). The results are not indicative of parochial vicarious ostracism. The interaction between gender and the exclusion condition was not significant, p = .34. Thus, vicarious ostracism in Study 2 was not a function of gender group membership. Distress was only a function of witnessing the exclusion of women.

Moderated Moderation to Test H1 and H2
H1 and H2 were tested as in Study 1. As can be seen in Table 5, a significant model was observed that tested three main effects, three two-way interactions, and one threeway interaction, explaining 29% of the variance in distress. Specifically, witnessing the exclusion of women increased distress and the association between gender collective narcissism and distress was significant and positive. These main effects were qualified by a significant two-way interaction between gender and exclusion, and the three-way interaction between exclusion, gender, and gender collective narcissism. Figure 2 depicts the three-way interaction. Table 4 presents the simple slopes for men and women in each condition.
Consistent with H1, gender collective narcissism was positively associated with distress, especially among women who observed the exclusion of women. It was also associated with distress among women in the control condition, but the relationship was half the size. With respect to H2, gender collective narcissism was negatively associated with distress when men observed the exclusion of women (p = .06), and positively but not significantly associated with distress among men in the control condition (p = .28). To specify H1 and H2, the same analyses were performed with gender identification as a continuous moderator. Those analyses revealed significant main effects of exclusion (b = 1.03, SE = .12, t(377) = 8.70, p < .001), and gender (b = -.34, SE = .14, t(377) = -2.55, p = .01) and a two-way interaction between gender and gender identification on distress (b = -.41, SE = .17, t(377) = -.39, p = .02). Gender identification was negatively associated with distress across conditions among men (b = -.37, SE = .09, t(377) = 4.20, p < .001), but not associated with distress among women (b = .06, SE = .08, t(377) = 0.78, p = .44) (codes available at https:// osf. io/ 83d45/). In other words, only collective narcissism, and not gender ingroup identification, predicted parochial vicarious ostracism -feeling ostracized and distress at exclusion only of the gender ingroup but not when witnessing exclusion of the gender outgroup.
Additionally, the same analyses were performed using individual narcissism as a moderator. Separate analyses were conducted with vulnerable narcissism and grandiose narcissism as the narcissism literature points to the complexity and multifaceted nature of this concept (e.g., Sedikides, 2021). The analyses revealed only the positive association between vulnerable narcissism and distress and the main effect of exclusion. Neither form of individual narcissism moderated the effects of gender and exclusion on distress (codes available at https:// osf. io/ 83d45/). In other words, collective but not individual narcissism predicts parochial vicarious ostracism.
The results of Study 2 replicated the findings of Study 1 using a more internally valid manipulation of exclusion, demonstrating that, as expected, gender collective narcissism predicts greater distress at the exclusion of women among women but not among men. Nevertheless, Study 2 was also performed on a convenience sample and continuous moderators were assessed after the experimental manipulation in both studies. Study 1 and 2 did not explore whether men react to the exclusion of their gender ingroup vs. outgroup in the same way that women do. Thus, Study 3 was conducted to correct for the shortcomings of Study 1 and 2 by testing H3 and H4 in a nationally representative sample of Polish men and women.

Study 3
For Study 3, the hypotheses, design, and analytic strategy were pre-registered (https:// aspre dicted. org/ b6pi9. pdf). Study 3 aimed to provide an exact and conceptual replication of the findings from Study 1 and 2. Study 3 also tested whether the hypothesized effects generalize beyond one measure of negative emotions and therefore is not affected by how distress was assessed. In addition, Study 3 aimed to test H3 and H4, which predicted that gender collective narcissism among men would be associated with greater distress when men witness the exclusion of other men by women.
This prediction was important to test as the collective narcissism theory indicates that people may endorse collective narcissism with respect to any group they belong to with the same consequences for intergroup relations (Golec de Zavala, 2011;Golec de Zavala & Lantos, 2020;Golec de Zavala et al., 2009. Thus, the exaggerated reactions to the gender ingroup's exclusion should be symmetrical among women and men who endorse gender collective narcissism. Gender collective narcissism should predict the same vicarious reactions to exclusion among women and men even though women experience exclusion as a group more often than men.

Participants
Participants were 688 Polish adults (344 women and 339 men; age ranged from 18 to 84 years, M = 45.66, SD = 14.32) recruited by Ariadna Research Panel (ariadna.pl). The sample was nationally representative with respect to age, gender, education, place of residence, and income. Only participants who were not familiar with the Cyberball paradigm could participate in the study. Participants who did not answer the attention check questions correctly (e.g., 'Please choose answer 'Agree'') were automatically excluded from further participation in the study. Data were collected only from people who reported a binary gender identity. All measures were screened for outliers (3 SDs from the mean as per the pre-registered criteria). Data from 5 participants who did not meet this criterion for the measure of negative affect were removed from analyses that included this variable.
Negative affect was assessed using the 10 negative items of the Positive and Negative Affect Scale (PANAS, Watson et al., 1988), which has been used and validated in previous studies in Poland (Golec de Zavala, 2019), α = .92.

Procedure
The study followed a 3 (exclusion: control vs. exclusion of women by men vs. exclusion of men by women) × 2 (gender: men vs. women) design. Men and women were randomly assigned to exclusion conditions by the survey platform. The procedure was the same as in Study 2, with an additional exclusion condition in which a team of women excluded a team of men in the ball-tossing game. Demographic variables, gender collective narcissism, and gender identification were assessed before the experimental manipulation. We used an adapted intergroup Cyberball paradigm (Hase et al., 2021). In the exclusion of women condition (n = 221), participants observed women receiving 10 percent of the first 30 ball-passes at the beginning of the game and then men exchanged the ball tosses between themselves excluding women from the game. In the exclusion of men condition (n = 223), participants observed men receiving 10 percent of the first 30 ball-passes and then being excluded by women who exchanged the ball tosses between themselves for the rest of the game. In the control condition (n = 234), participants observed teams of men and women exchanging an equal number of ball-passes throughout the game.
After the game, participants responded to several manipulation check questions. Next, distress was assessed, as in Study 1 and 2, followed by negative affect. The scales were presented in a random order with the order of items within each scale randomized for each participant. All items were scored on a 7-point scale ranging from 1 (completely disagree) to 7 (completely agree). At the end of the study, participants were probed for guessing (nobody guessed the hypothesis or reported suspicion about the study), debriefed, and thanked.

Study 3 Results
Means, standard deviations, and zero-order correlations among all variables in Study 3 are presented in Table 6. Gender collective narcissism and gender identification were positively correlated. Gender collective narcissism, but not gender identification, was positively associated with distress and support for the women's strike.
In order to check that the manipulation of the exclusion of women was effective, a univariate 3 × 2 ANOVA was performed testing for the effect of exclusion on the manipulation check questions. As shown in Table 7, there was no significant main effect for exclusion condition. The twoway interaction between gender and exclusion condition was significant for perceived percentage of ball throws and perceived exclusion. The main effect of exclusion condition on perceived exclusion was qualified by a significant two-way interaction with gender. Differences between men and women were not significant in the control condition for percentage of ball throws (p = .06) or perceived exclusion (p = .09). Otherwise, the means were significant at p < .001. Men perceived less ball-throws thrown to men by women and felt more excluded in the exclusion of men condition. Analogously, women perceived less ball-throws thrown to women and felt more excluded in the exclusion of women condition.
The results of a 3 × 2 ANOVA testing for vicarious ostracism and parochial vicarious ostracism indicated the presence of parochial vicarious ostracism in Study 3. The main effect of exclusion was significant for distress, F(2, 682) = 34.36, p < .001, ω 2 = .09, and negative affect, F(2, 677) = 11.60, p < .001, ω 2 = .03. This main effect was qualified by a significant two-way interaction with gender for distress, F(2, 682) = 26.80, p < .001, ω 2 = .07, and negative affect, F(2, 682) = 4.20, p = .015, ω 2 = .01. As can be seen in Table 8, women felt more distressed when witnessing the exclusion of women in comparison to the control condition and witnessing the exclusion of men, and to both conditions combined (linear contrast, Difference = -.42; SE = .05, p < .001). Men felt more distressed when witnessing the exclusion of men in comparison to the control condition and witnessing the exclusion of women, and to both conditions combined (linear contrast,  Difference = -.45; SE = .07, p < .001). As can be seen in Table 8, a similar pattern can be observed for negative affect. Women felt more negative affect when witnessing the exclusion of women in comparison to the control condition but not compared to witnessing the exclusion of men. Nevertheless, they reported more negative affect in the exclusion of women condition compared to the remaining two conditions combined (linear contrast, Difference = .18; SE = .05, p < .001). Men felt more negative affect when witnessing the exclusion of men in comparison to the control condition but not compared to witnessing the exclusion of women. Men reported more negative affect when witnessing the exclusion of men compared to the remaining two conditions combined (linear contrast, Difference = .12; SE = .04, p = .007).

Moderated Moderation to Test H1 and H2
H1-H4 were tested with a moderated moderation analysis using a hierarchical multiple regression. A three-way interaction between gender collective narcissism (centered), gender (0 = women, 1 = men), and exclusion (dummy coded, -1 = exclusion of men, 0 = control, 1 = exclusion of women) in predicting distress was expected (Hayes, 2022; Model 3 using indicator method for the multi-categorical independent factor). A significant model testing four main effects, five two-way interactions, and two three-way interactions explained about 19% of the variance in distress. The main effects for gender, gender collective narcissism, and exclusion were qualified by significant two-way interactions between exclusion and gender collective narcissism. As shown in Table 9, the two-way interactions were qualified by a significant three-way interaction between gender, exclusion, and gender collective narcissism. Figure 3 illustrates the interactions. Table 10 presents simple slopes for men and women in each condition. As the results in Table 10 indicate, consistent with H1, gender collective narcissism was positively associated with distress among women who witnessed the exclusion of women. With reference to H2, gender collective narcissism was negatively associated with distress among men who witnessed the exclusion of women. However, this association was not statistically significant. These results replicate the findings from Study 1 and 2. Consistent with H3, gender collective narcissism was positively associated with distress among men who witnessed the exclusion of men. With reference to H4, gender collective narcissism was negatively associated with distress among women who witnessed the exclusion of men. However, this association was not statistically significant. The associations between gender collective narcissism and distress at the exclusion of the ingroup were almost exactly symmetrical between men and women.
To test H1-H4 using an alternative assessment of distress, the same analysis was performed with negative affect as the outcome variable. As can be seen in Table 11, a significant model testing for the hypothesized three-way interaction explained about 11% of variance in negative affect. As shown in Fig. 3 and Table 10, and consistent with H1, gender collective narcissism was associated with negative affect among women who witnessed the exclusion of women. In line with H3, gender collective narcissism was associated with negative affect among men who witnessed the exclusion of men. Contrary to H2 and H4, the association between gender collective narcissism and negative affect did not reverse direction when men and women witnessed the exclusion of the gender outgroup and were nonsignificant. Thus, gender collective narcissism predicted negative affect among men and women who witnessed the exclusion of their gender ingroup, but not among participants who witnessed the exclusion of their gender outgroup.
To specify H1-H2, the same analyses were performed with gender identification as a continuous moderator. Those analyses revealed no significant three-way interaction for distress, F(2, 676) = 1.73, p = .18, or negative affect, F(2, 671) = 2.73, p = .07. Thus, unlike gender collective narcissism, gender identification was not associated with distress at the gender ingroup's exclusion among men and women in Study 3.

General Discussion
The present experiments aimed to explain why men do not universally support women who contest the exclusion of women from public life, and why some women may be more mobilized to challenge gender discrimination than others. Three experiments examined women's and men's distress when witnessing the exclusion of women as a function of gender collective narcissism. In addition, one study examined how men and women who endorse gender collective narcissism also react to the exclusion of men. Positive associations were expected between gender collective narcissism and distress at the exclusion of the gender ingroup relative to exclusion of the outgroup. The results of all three studies converged to support the conclusion that, among men and women alike, gender collective narcissism predicts parochial vicarious ostracism by moderating the distress experienced by women and men witnessing the exclusion of the gender ingroup and outgroup, i.e., greater distress when participants witness the exclusion of the gender ingroup rather than the gender outgroup. The present research did not find consistent support for vicarious ostracism, i.e., greater distress at witnessing exclusion vs. control (Wesselmann et al., 2013), or parochial vicarious ostracism, i.e., greater distress at witnessing the ingroup vs. the outgroup exclusion) in the context of gender exclusion. Instead, the disparity in the way men and women react to witnessing their gender ingroup vs. outgroup exclusion was consistently associated with gender collective narcissism. Moreover, this effect was specific to gender collective narcissism and did not hold for gender identification (or any other aspect of it) or individual narcissism. While the findings regarding the role of gender collective narcissism are consistent across all three studies, the findings regarding the predictions for gender identification were not.
In line with H1 and H3, in all studies, the association between gender collective narcissism and distress was significant and positive after witnessing the exclusion of the gender ingroup. In Study 1, distress was higher among women (vs. men) who endorsed gender collective narcissism when they watched the female MP silenced by the male Chairman during the Polish Parliament session. In contrast, the association between gender collective narcissism and distress among men was not significant. The same pattern of results was replicated in Study 2 and 3, in which the exclusion of women was manipulated using the context-free, abstract, ball-tossing game. Those studies sacrificed the ecological validity of Study 1 to increase the internal validity of the experimental manipulation of the exclusion of women. Thus, in all studies, while watching the exclusion of a woman by a man (vs. control), female collective narcissists became more distressed, whereas male collective narcissists' distress levels did not change.
Study 3 replicated those results using an alternative assessment of distress. It also showed that gender collective narcissism predicted parochial vicarious ostracism among women and men alike. While the relationship between gender collective narcissism and distress was positive and significant for women witnessing the exclusion of other women, this association was also positive, significant and of nearly the same size for men witnessing the exclusion of other men. The association was not significant when women witnessed the exclusion of men and when men witnessed the exclusion of women.

Gender Collective Narcissism Predicts Parochial Vicarious Ostracism
Only one of the present studies revealed a pattern of findings consistent with vicarious ostracism, i.e., feeling distressed by witnessing the exclusion of other people. Instead, across all studies, men were less distressed by the exclusion of women when they endorsed gender collective narcissism. An analogous association was observed among women in Study 3. Women who endorsed gender collective narcissism reported less distressed at the exclusion of men.
The present results extend previous findings regarding vicarious ostracism (Wesselmann et al., 2013) indicating that vicarious ostracism in a gender intergroup context is more likely among people who endorse gender collective narcissism. However, at higher levels of gender collective narcissism vicarious ostracism becomes parochial, whereby distress is experienced only for the gender ingroup but not for the gender outgroup. The present findings also indicate that perceived similarity to or empathy with excluded others is not always necessary to experience parochial vicarious ostracism. Even exclusion of unknown, anonymous members of the gender ingroup (e.g., represented only by abstract avatars in the virtual ball tossing game) is experienced as personal exclusion and causes distress among people who endorse gender collective narcissism.
Interestingly, the role of gender collective narcissism in intensifying distress at the exclusion of the gender ingroup parallels the role of individual narcissism which enhances the negative effects of interpersonal exclusion (Blinkhorn et al., 2021;Twenge & Campbell, 2003). However, the present findings indicate that gender collective narcissism not individual narcissism, enhances the negative effects of exclusion of the gender ingroup. These results are in line with previous findings that collective narcissism parallels the role of individual narcissism on the social level of the self and that only collective, but not individual narcissism, is associated with emotions, attitudes and behavioral intentions in various intergroup contexts (Golec de Zavala et al., 2009, 2022Golec de Zavala, 2011). The present studies show that the same is found in the gender intergroup context. Gender collective narcissism, not individual narcissism of men and women, predicts parochial vicarious ostracism. The present findings also help clarify why previous research might have produced inconsistent findings with reference to the effects of attributing personal exclusion to one's group membership. One previous study demonstrated that distress was aggravated when women who were excluded in the Cyberball game attributed their exclusion to their group membership (Schaafsma & Williams, 2012). However, anotther study showed that members of different disadvantaged groups (i.e., an ethnic minority) who were excluded in the Cyberball game and attributed their exclusion to their ethnic group membership felt less distressed by exclusion (Masten et al., 2011). The inconsistency might have been produced not only by different intergroup contexts but also by previous studies not taking participants' collective narcissism into account.
The present results also clarify the seemingly contradictory findings regarding the role of ingroup identification in the face of social identity threat (Eliezer et al., 2010;Fischer & Holz, 2007;McCoy & Major, 2003). Witnessing exclusion of a member of the gender ingroup did not increase gender identification or gender collective narcissism as might be predicted based on the Rejection Identification Model. In addition, gender ingroup identification did not affect participants' reactions to witnessing the exclusion of ingroup members. Instead, gender collective narcissism was a factor that enhanced group members' distress in the face of the gender ingroup's exclusion. Exclusion is a threat to the ingroup's image (Branscombe et al., 1999a, b), to which collective narcissists are particularly sensitive (Golec de Zavala et al., 2016), and to which they react with hostility and aggression (Golec de Zavala et al., 2013Guerra et al., 2022;Hase et al., 2021). The present results suggest that studies that do not differentiate collective narcissism from other aspects of ingroup identification when researching reactions to intergroup exclusion are likely to produce seemingly contradictory findings.

Limitations and Further Research Directions
The present results were obtained in a Polish national context which raises the question as to whether those findings generalize to different national contexts. Future studies would do well to test similar hypotheses in different countries, also those in which the status of men and women is more equal than in Poland. Women's rights face new threats worldwide with ultraconservative populism advancing gender inequality (Graff & Korolczuk, 2021). This is particularly visible in Poland were women were stripped of their reproductive rights in October 2020. However, similar changes are impeding in national contexts where women have enjoyed more equal status as illustrated by the recent overruling of the Roe vs. Wade decision that has guaranteed women's reproductive rights in the United States.  Regardless of the national context, women who endorse gender collective narcissism and witness institutionalized exclusion, silencing, and the oppression of other women are likely to feel distressed, and also are likely to fight back (Golec de Zavala & Keenan, 2022a, b). Future studies would do well to examine whether gender collective narcissism among women predicts increased motivation to engage in collective action after witnessing the exclusion of women. Studies should also examine the role of gender collective narcissism among men in shaping attitudes towards allyship with women.
Future studies would also do well to explore the effect of positive prosocial emotions experienced during collective action opposing exclusion (Pereira et al., 2022) on the association between gender collective narcissism and intergroup antagonism among women and rejection of allyship among men. The association between collective narcissism and intergroup antagonism is reduced by experiencing prosocial emotions . Such findings suggest that efforts to emphasize compassion, feeling touched by and connected to the social world beyond the self through collective action may reduce the negative intergroup effects of gender collective narcissism and motivate women's involvement in collective action opposing gender discrimination (Golec de Zavala & Keenan, 2022a).

Practice Implications
The present results align with previous findings suggesting that gender collective narcissism may be an obstacle to the allyship and solidarity between men and women in pursuit of gender equality and social justice. Previous studies showed gender collective narcissism among men predicts sexism (Golec de Zavala & Bierwiaczonek, 2021) and lack of solidarity with protesting women (Górska et al., 2020) in Poland. Moreover, gender collective narcissism is associated with opposing attitudes towards gender equality among men and women; specifically, gender collective narcissism is associated with egalitarianism among women but the legitimization of gender inequality among men. Gender collective narcissism is also linked to support for the All-Poland Women's Strike among women but refusal to support the strike among men. However, among men and women, gender collective narcissism predicts perceiving gender relations as a zero-sum conflict, in which one can only win by coercively advancing the interests of the gender ingroup against the interests of the outgroup (Golec de Zavala & Keenan, 2022b). Such findings suggest that the greater distress experienced when witnessing the gender ingroup's exclusion among individuals with higher gender collective narcissism may motivate men and women to endorse beliefs and attitudes that impede men's and women's allyship in pursuit of gender equality. Constructive systemic change towards gender equality is only possible when men and women cooperate to achieve social justice. However, gender collective narcissists among men may not be interested in advancing fairness and gender equality because, as the present results elucidate, they are not distressed by the discrimination of women. Instead, they are distressed at the prospect that men as a group may face marginalization and exclusion. Thus, men are likely to fail in allyship with women as long as they perceive gender equality as an intergroup threat, and they are more likely to perceive gender equality as a threat when they endorse gender collective narcissism . Increasing awareness of and discouraging narcissistic definitions of gender identity among men may assist in building the way forward towards gender equality.
The finding that gender collective narcissism is associated with preoccupation with the gender ingroup's exclusion, whether it is real (as is the case of women) or mostly imagined and feared (as is the case of men), explains not only why men are not likely to support collective action for gender equality, but also why women who engage in collective action for gender equality may be hesitant to accept allyship from men (Cheng et al., 2019;Park et al., 2022). Collective narcissism is negatively associated with trust (Golec de Zavala et al., 2009), but positively associated with paranoid and conspiratorial thinking (Golec de Zavala & Cichocka, 2012;Golec de Zavala et al., 2022). Trustworthiness is a crucial characteristic sought in allies from advantaged groups (Park et al., 2022). Considering these findings, it would be a painful paradox if gender collective narcissism motivated women toward collective action for gender equality (Golec de Zavala & Keenan, 2022a, b), but would, at the same time, undermine their trust in men as allies.

Conclusion
Across three studies, gender collective narcissism predicted parochial vicarious ostracism in the face of gender-based exclusion, i.e., feeling distressed only at the exclusion of the gender ingroup but not the gender outgroup, which may be an obstacle to allyship between men and women and the pursuit of gender equality. Among women, gender collective narcissism was associated with distress and negative affect when witnessing the exclusion of women but not when witnessing the exclusion of men. Among men, gender collective narcissism was associated with distress and negative affect when witnessing the exclusion of men but not when witnessing the exclusion of women. These findings suggest that discouraging narcissistic definitions of gender identity, especially among men, may be necessary for gender allyship and the pursuit of gender equality.