1 Introduction

Experiences of bullying are very frequent among youth with group-based minority identities (i.e., so-called bias-based bullying; e.g., ethnic minority youth, or sexual minority youth; Earnshaw et al., 2018; Russell et al., 2012). One of the most common forms of bias-based bullying is homophobic bullying: bullying based on actual or perceived lesbian, gay, bisexual, queer, or gender non-conforming identity (e.g., Russell et al., 2012). Homophobic bullying affects all students regardless of their sexual orientation, gender identity, gender expressions, or characteristics (Espelage et al., 2018; Kosciw et al., 2009) and has several detrimental consequences (e.g., depression and lower levels of school engagement; Berlan et al., 2010; Kosciw et al., 2009). Indeed, research demonstrated that besides sexual minority youth, homophobic bullying might be directed to heterosexual youth who may somehow be perceived as being different in terms of traditional gender role expectations (e.g., Poteat & Espelage, 2005).

Research focusing on how to reduce bullying in the school context has moved beyond an interpersonal approach focusing on the dyadic dynamics between youth who bully and youth who are victimized, to an intergroup approach focusing on the intergroup dynamics of those who are victimized, those who bully and the bystanders (i.e., those who witness bullying incidents; Palmer & Abbott, 2017).Bystanders are often present in bullying episodes (Jones & Rutland, 2019; Sutton & Smith, 1999), and have a relevant role in facilitating or inhibiting bullying (Salmivalli, 2010). When bystanders reinforce bullying and fail to defend victims, bullying prevalence is higher (e.g., Salmivalli et al., 2011; Thornberg & Wänström, 2018). On the contrary, research shows when bystanders challenge bullying, they can reduce it (Garandeau et al., 2023; Palmer & Abbott, 2017), suggesting that bystanders are effective at reducing bias-based bullying (Abbott & Cameron, 2014; Mulvey et al., 2016; Palmer & Abbott, 2017).

However, most research on how to promote bystanders’ assertive interventions (i.e., to help the victim or stop the perpetrator) is correlational (e.g., Abbott & Cameron, 2014), and it is not possible to determine what strategies are effective to trigger their intervention in bullying episodes. Additionally, most research focused on “traditional” bullying, and less is known about the factors that can promote assertive interventions in the specific context of homophobic bullying in schools. The present study extends previous research by examining, experimentally, the impact of different forms of common inclusive identities (i.e., one-group and dual-identity representations) on youth bystanders’ behavioral intentions to help homophobic bullying victims. Considering that previous research shows that as children and adolescents get older, they become less likely to defend those who are victimized (Palmer & Abbott, 2017), we considered the developmental period (i.e., middle adolescence to late adolescence) in which these helping behaviors occur. Understanding this decline and how to improve middle and late adolescents’ intentions to help homophobic bullying victims is important when creating effective interventions to be implemented in schools. We also considered potential sex differences, considering that research suggests that male adolescents exhibit more negative attitudes towards sexual minorities (e.g., Costa & Davies, 2012), while female adolescents tend to score higher in defending behaviors during bullying incidents than their male counterparts (e.g., António et al., 2022; Pouwels et al., 2016; Pozzoli & Gini, 2012).

1.1 Common identities and intergroup helping responses

The endorsement of common identities, inclusive of both ingroup and outgroup, is a powerful strategy to improve intergroup relations and reduce prejudice (see Gaertner et al., 2016 for a review). The Common Ingroup Identity Model (CIIM) proposes that intergroup bias can be reduced by inducing members of different groups to recategorize as members of the same more inclusive group. Recategorization can take different forms: a common inclusive identity (i.e., one-group) or a dual-identity, which involves the simultaneous activation of a common identity and original subgroup identities (Gaertner et al., 1993). The effectiveness of common identities for improving intergroup relations depends on the group's status (Gaertner et al., 2016; Hehman et al., 2012). Research shows that majority group members usually prefer one-group representations that focus on commonalities and reduce the emphasis on subgroup identification, whereas minority group members prefer a dual-identity representation that recognizes group distinctiveness and disparities (e.g., Dovidio et al., 2001; Hehman et al., 2012). Conversely, dual-identity, relative to one-group, is more effective in triggering solidarity-based collective action among majority groups (e.g., Banfield & Dovidio, 2013).

Despite these differential effects, recategorization, as proposed by the CIIM, is an effective strategy to reduce bias and prejudice across different ethnic (e.g., Kunst et al., 2015), political (e.g., Riek et al., 2010) and age groups (e.g., young children, Cameron et al., 2006; Guerra et al., 2010; and adolescents, e.g., Gaertner et al., 1996). Besides reducing prejudice, research shows that inducing common inclusive identities also increases adult bystanders’ intentions to help those who are victimized during emergency contexts (Levine et al., 2005). One study, involving college students self-identified as Manchester United FC supporters, revealed that when a common identity was made salient (i.e., wearing an ingroup team shirt), participants were more likely to help the person who was victimized, than when wearing a rival team shirt (Levine et al., 2005). Thus, sharing a common identity increases the likelihood of helping. Consistent with the findings with adults, research, conducted with heterosexual youth, showed that adolescent’s endorsement of a common identity (i.e., thinking of both heterosexual and gay/lesbian students as one-group) was positively related to their behavioral intentions to help homophobic bullying victims (António et al., 2018).

2 The current study

In the current study, we extended previous correlational studies (e.g., António et al., 2018) examining, experimentally, the effects of both forms of common identities, one-group and dual-identity (vs. a control condition) on intentions of helping the victims of homophobic bullying. Specifically, we examined if inducing one-group and dual-identity representations among heterosexual adolescents (12–19 years) increased their behavioral intentions to help homophobic bullying victims (vs. control condition where no common ingroup identity is salient). Building on previous research, in the one-group condition a common inclusive identity is emphasized; the dual-identity condition involves the simultaneous activation of a common identity and original subgroup identities; and in the control condition, no common ingroup identity is salient (e.g., Shnabel et al., 2013; Ufkes et al., 2016).

Based on previous research showing that both forms of inclusive identities are effective to promote positive intergroup relations (Gaertner et al., 2016) and have been related to bystanders’ helping behavioral intentions (António et al., 2018), we predicted that participants in the one-group and dual-identity conditions (vs. control condition) would show more behavioral intentions to help the victims (H1). Considering that previous research also showed differential effects of one-group and dual-identity representations for majority and minority groups (e.g., Hehman et al., 2012), we also examined if inducing a common identity that does not make salient the different sexual orientations of participants (one-group) is more effective in increasing helping behavioral intentions among majority, heterosexual, adolescents, than dual-identity, that involves the salience of the original subgroups (H1a).

Women and girls are usually more likely to engage in defending and supportive behaviors than men and boys (e.g., Pozzoli & Gini, 2012). Based on these findings, we predicted that female adolescents (vs. male), in the one-group and dual-identity conditions (versus. control condition) would show more helping behavioral intentions (H2). Finally, we also explored possible age differences. Before puberty, sexual maturity, and sexual identity formation, adolescents may use homophobic language without fully comprehending its meaning. This language usage is primarily aimed at addressing deviations from social norms among their peers (Falomir-Pichastor & Mugny, 2009; Plummer, 2001). As adolescents grow older, their sexual identity becomes more defined, and the peer-group culture becomes saturated with homophobia. At this stage, adolescents recognize the meaning behind homophobic language and may use it with homophobic intent (Plummer, 2001). Moreover, homophobic bullying and name-calling spikes during adolescence (e.g., Russell et al., 2012). Thus, it is reasonable to expect that older adolescents may be less inclined to offer help in homophobic bullying situations as it may be less socially accepted to defend a victim of this form of bullying. Building on this reasoning but considering the lack of previous research examining age differences on the examined variables across adolescence, we did not establish directional hypotheses for age and explored its potential moderator role. A better understanding of this developmental aspect is important to explore potential variations in bystanders' helping intentions across different age groups, contributing to a deeper understanding of how age may influence responses to homophobic bullying incidents.

3 Method

3.1 Participants and procedure

Participants were 185 Portuguese students (100 female) from Lisbon Metropolitan area, aged between 12 and 19 years (M = 15.43, SD = 1.69). Twenty nine percent were in middle school and 71% were in high school (10–12th years). Most participants identified as heterosexual (92%) and the remaining as gay/lesbian or bisexual, did not respond or declared having doubts regarding their sexual orientation. Since the outgroup target in this experiment was gay/lesbian adolescents, the final sample included only participants who self-identified as heterosexual (170 students; M = 15.44, SD = 1.71; 90 female). Participants were divided into two groups according to their age/development period: middle adolescence (< 16 years) and late adolescence (> 16 years). The sample size was determined based on schools and students’ availability/willingness to participate in the study. A sensitivity power analysis indicated that the final sample size was adequate to detect effects as small as d = 0.24 with 0.80 power (G*Power 3.1; Faul et al., 2007).

The survey was approved by the institutional Ethics Committee and conducted following the ethical standards of the American Psychological Association, the Declaration of Helsinki, and the European General Data Protection Regulation. Data were collected in two public schools, all participants provided previous parental consent and their agreement to participate. Participants were randomly assigned to one of three experimental conditions (control, one-group, dual-identity) and completed a paper and pencil questionnaire in classrooms. The questionnaire presented one of three different scenarios, fabricated by the research team, depending on the experimental condition. All scenarios (matching participants’ sex) described a verbal and physical aggressive event toward a student that happened some years ago during a soccer game in their school. After that, the manipulation of common identities was introduced. In the one-group condition the victim of the aggression was described as a member of the same inclusive group (i.e., “girl/boy from your school”). In the dual-identity condition the victim was presented as a member of the same inclusive group (school) highlighting simultaneously the sexual orientation (“girl/boy from your school who was with her/his girlfriend/boyfriend”). In the control condition, the victim was described as someone from a different group (rival school).

After reading the scenario, participants were asked to imagine that they were watching the game and to fill out a questionnaire with all the measures of interestFootnote 1 After completion of the materials, all students received a written debriefing.

3.2 Measures

3.2.1 Helping behavioral intentions

Participants indicated their intention to engage in 4 helping bystanders’ behavioral intentionsFootnote 2 (Palmer & Cameron, 2010), on a 5-point scale (1 = never do to 5 = always do; e.g., “I would try and make student B feel better”; “I would tell person A not to say nasty things”, α = 0.68). The original measure included 10 bystanders’ behavioral intentions but, in this study, we focused only on helping bystander intentions.

4 Results

Descriptives are presented in Table 1.

Table 1 Means, standard deviations, main effects and interaction effects by condition (N = 170)
Full size table

To test H1 and H1a, we conducted an Analysis of Variance (ANOVA) on helping behavioral intentions, F(2, 158) = 6.13, p = .003, η2p = 0.72. We computed two orthogonal contrasts: control versus one-group and dual-identity conditions pooled (C1), and one-group versus dual-identity condition (C2). Sex and age were also entered as factors to explore its moderator role. Contrary to H1, orthogonal contrasts showed that one-group and dual-identity conditions pooled did not differ significantly from the control condition, t(157) = −0.20, p = .85, d = 0.03, 95% CI [−0.28, 0.23]. Contrary to H1a, orthogonal contrasts showed that helping behavioral intentions were higher in the dual-identity condition (M = 4.11, SD = 0.83) than in the one-group condition (M = 3.52, SD = 0.92), t(157) = −3.51, p < .001, d = 0.67, 95% CI [−0.84, −0.23].

Considering the non-significant differences between one-group and dual-identity conditions pooled and the control condition, we conducted additional exploratory analyses to better understand this effect. Two additional contrasts were computed to compare separately one-group versus control and dual-identity versus control condition. Results showed that helping behavioral intentions were higher in the dual-identity condition (M = 4.11, SD = 0.83) than in the control condition (M = 3.79, SD = 0.76), t(167) = 2.08, p = .04, d = 0.39, 95% CI [0.02, 0.74]. However, the one-group and control conditions did not differ significantly, t(167) = −1.69, p = .09, d = 0.32, 95% CI [−0.70, 0.05]. In sum, participants in the dual-identity condition showed higher helping behavioral intentions relative to those in the control and one-group conditions.

The interaction effects with sex and age group were also significant. Contrary to H2, results showed that female adolescents, in the one-group and dual-identity conditions (vs. control condition) did not show more helping behavioral intentions as predicted, t(83) = −0.27, p = .79, d = 0.06, 95% CI [−0.33, −0.25]; that is, female adolescents, in the one-group and dual-identity conditions (vs. control condition) did not show more helping behavioral intentions as predicted. Additionally, female adolescents in the dual-identity condition showed higher helping behavioral intentions (M = 4.32, SD = 0.54) than those in the one-group condition (M = 3.67, SD = 0.82), t(83) = −3.58, p < .001, d = 0.94, 95% CI [−0.96, −0.27]. Overall, these results suggest that the effects on inclusive identities were moderated by sex.

Finally, results also revealed a marginal 3-way interaction between the experimental condition, age, and sex, F(2,158) = 2.89, p = .059, η2p = 0.035. Orthogonal contrasts showed that bystanders’ helping behavioral intentions in the one-group and dual-identity conditions did not differ significantly from the control condition for both male (tyounger (34) = −0.50, p = .62, d = 0.17, 95% CI [−0.80, 0.49] and tolder (38) = 0.45, p = 0.65, d = 0.15, 95% CI [-0.46, 0.73]) and female participants (tyounger (43) = −0.26, p = .80, d = 0.08, 95% CI [−0.45, 0.34] and tolder (39) = −0.12, p = 0.90, d = −0.04, 95% CI [−0.46, 0.41]). However, orthogonal contrasts comparing dual-identity vs. one-group (C2) showed significant differences for younger female participants, who showed higher helping behavioral intentions in the dual-identity condition (M = 4.53, SD = 0.40), than in the one-group condition (M = 3.65, SD = 0.96), t(43) = −3.75, p < .001, d = 1.19, 95% CI [−1.34, −0.41]. For male participants the pattern of findings was the opposite: helping behavioral intentions were higher for older participants in the dual-identity condition (M = 3.86, SD = 1.03), compared to the one-group condition (M = 2.96, SD = 0.96), t(38) = −2.32, p = .03, d = 0.90, 95% CI [−1.69, −0.12]. Overall, these effects suggest that sex and age moderated the effects of different forms of inclusive identities.

Additionally, results revealed significant main effects of sex, F(1,158) = 8.36, p = .004, η2p = 0.050; and age, F(1,158) = 9.78, p = .002, η2p = 0.058. Female participants showed more helping behavioral intentions (M = 4.03, SD = 0.71) than male participants (M = 3.60, SD = 0.96). Regardless of condition, helping bystanders’ behavioral intentions were greater among younger (M = 4.05, SD = 0.79) than older participants (M = 3.60, SD = 0.88).

5 Discussion

This study examined whether inclusive identities (i.e., one-group and dual-identity) can positively impact bystanders’ behavioral intentions to help homophobic bullying victims. Additionally, we also examined potential sex differences and explored the developmental trends in helping behaviors. Overall, our results showed that inducing a dual-identity representation that simultaneously emphasizes a common ingroup and subgroup identities resulted in more positive outcomes relative to inducing a single one-group that does not highlight different subgroups.

Contrary to our hypotheses (H1), the experimentally induced common inclusive identities (one-group and dual-identity) did not significantly increase helping behavioral intentions among heterosexual adolescents compared to the control condition. Indeed, when examining the pooled effect of both forms of recategorization together no differences were found for participants helping intentions. However, our results showed significant difference between the two forms of common identity representations, in line with previous evidence suggesting that one-group and dual-identity effects differ (H1a). Indeed, adolescent participants in the dual-identity condition demonstrated more helping behavioral intentions compared to those in the one-group condition.

Additional exploratory analyses also showed that dual-identity resulted in more helping behavioral intentions relative to the control condition. This finding aligns with previous research showing dual-identity's effectiveness in promoting solidarity-based collective action among majority groups (e.g., Banfield & Dovidio, 2013). Helping a victim of homophobic bullying may be considered as a form of solidarity behavior towards a minority group (i.e., a gay or lesbian student), without threatening a valued subgroup identity (i.e., heterosexual). Future studies are needed to better understand the efficacy of both forms of common identity representations, examining if different underlying mechanisms account for their effects on helping behavioral intentions. In addition, given that members of minority groups tend to prefer a dual-identity representation (Dovidio et al., 2001), future research could also focus on the responses of sexual minority youth bystanders to bullying episodes, examining the effects of both forms of recategorization for minorities and majorities. This will allow us to understand if the impact of one-group and dual-identity on bystanders’ helping behavioral intentions generalizes for majority and minority groups.

Our results also showed significant interactions between the experimental manipulation of inclusive identities, age, and sex, highlighting the importance of considering those factors when examining bystanders' helping intentions across different age groups, female and male adolescents. Overall, younger female revealed higher behavioral intentions to help in the dual-identity condition compared to the one-group condition. Similarly, older male participants in the dual-identity condition demonstrated higher helping behavioral intentions compared to the one-group condition. These results suggest that the impact of different common identity representations may vary across different age and sex groups. However, these findings should be interpreted with caution considering the small number of participants per cell. Although sensitivity power analysis indicated that our sample size was adequate to detect small effects, these were for the main effects of condition and not for two- and three-way interactions. Thus, it is crucial that future research replicates these findings with well powered samples. Additionally, further research is needed to better understand the factors that influence bystander helping behaviors in specific subgroups and age categories, particularly in the context of homophobic bullying.

In general, the differential findings for male and female adolescents were consistent with previous research (e.g., Pozzoli & Gini, 2012), showing that females generally display more behavioral intentions to help than males. Additionally, helping behavioral intentions were higher among younger participants compared to older ones, regardless of the experimental condition. This finding aligns with previous research that shows a decline in helping behaviors as children and adolescents grow older (Palmer & Abbott, 2017; Thornberg et al., 2020). It is important to take these developmental differences into account when designing interventions and programs to encourage bystander’ helping behaviors in the context of homophobic bullying.

Additionally, it is important to recognize the impact of homophobic bullying within the broader spectrum of sexual identity exploration during adolescence. Teenagers often undergo a process of understanding and developing their sexual identity, and experiences like witnessing or intervening in homophobic bullying can influence this development. Research shows that the fear of being perceived as gay or lesbian by association may prevent some heterosexual adolescents to engage in intentions of helping victims of homophobic bullying (António et al., 2018). Thus, it may be more socially acceptable to defend victims at younger ages (Ma et al., 2019), since that with age the influence of the peer group and the importance of the group functioning increase (Mulvey et al., 2016).

Overall, our experimental findings extend existing research showing the impact of common identities in increasing intergroup helping (e.g., Dovidio et al., 2017). Theoretically, these findings extend previous research in several ways. They contribute to the literature regarding the effects of common inclusive identities on a very prevalent intergroup context of bullying—homophobic bullying. In addition, by considering age and sex trends, this study highlights the importance of relying on a developmental social psychological approach to understand how children and adolescents develop their understanding of intergroup processes that may influence their responses to social situations, their intergroup attitudes, and their behavioral intentions (e.g., Gönültaş & Mulvey, 2020; Killen et al., 2013; Palmer et al., 2015).

6 Limitations and future directions

Despite the positive effects of common ingroup identities on adolescents’ helping intentions toward youth who are victimized in the form of homophobic bullying, it is still unclear why only dual-identity positively impacted bystanders’ helping intentions. This could be further examined in future studies exploring the underlying mechanisms that can account for this effect (e.g., intergroup anxiety; social contagion concerns).

The lack of effects of one-group representations could also be related to specific features of the manipulation scenario. The present study utilized fabricated scenarios, which may not fully capture the emotional and situational complexities that occur during real-life bullying incidents. In actual bullying situations, the interplay of various factors, such as peer pressure, fear of retaliation, and the presence of authority figures, can significantly influence bystanders' responses (Poteat & Vecho, 2015). Thus, the fabricated scenarios may have failed to capture the multiple factors involved in bullying situations real-life experiences. Thus, future studies could delve into these real-life dynamics, for instance, by using a field experiment or examining actual bullying incidents, to gain a more comprehensive understanding of bystander helping behavior in this specific context. Future studies could also manipulate a different common ingroup identity, as members of the same school might not trigger or evoke a sufficiently strong common (ingroup) identity as students might have experience a huge student heterogeneity in their own school, contrary to the dual-identity condition that is the only condition that clearly could be interpreted as bias-based bullying targeting a member of a minority group, and might therefore be considered as more severe and harmful, increasing a stronger sense of injustice, the sympathy for the victim, and the motivation to intervene.

Also, this study focused on two broad age groups and assessed broad age-related differences using a cross-sectional approach. Future studies could be conducted longitudinally to better capture the developmental trends in adolescent’s helping responses to bullying, and to identify potential avenues for its prevention.

As referred, the sample size may also have affected the significance of our effects, particularly the small number of participants per cell. The study was underpowered to test two-way and three-way interaction effects and these findings should be interpreted with caution.

It is also important to recognize that self-reported intentions may not accurately mirror the real bystander behaviors of young individuals in scenarios similar to the one presented. Importantly, in this study, we focused on behavioral intentions rather than actual observable behaviors. Although previous research revealed that behavioral intentions precede actual behaviors (e.g., Ajzen, 1991), other research reveals that directly measuring actual behavior has significant advantages (Crano et al., 2014), so it would be important to include measures that reflect actual bystanders’ behaviors (e.g., measuring bystander responses in real-time online bullying fictional incidents).

Another consideration is related to the school environments and teachers’ attitudes and beliefs. Research has shown that teachers’ attitudes and beliefs play a role in the perception of homophobic bullying and the promotion of strategies against it (e.g., Bacchini et al., 2020). Future research may look at which strategy (one-group or dual-identity) teachers may find more likely to promote more supportive and inclusive attitudes.

7 Practical implications

The findings suggest that enhancing the salience of a common ingroup identity among adolescents can be an effective strategy for building inclusive schools. This approach could be integrated into interventions to foster a sense of solidarity without threatening individual subgroup identities. The positive effects of Gay-straight alliances (GSAs) among heterosexual adolescents (e.g., Toomey & Russell, 2016) suggest that implementing such school-based programs could be an effective approach. GSAs can contribute to creating a more inclusive and less stigmatizing school environment, fostering assertive bystanders and promoting a sense of community among students.

The results also emphasized the need to consider age and sex differences when designing interventions. Future interventions should be tailored to specific age and sex groups to maximize their effectiveness in addressing homophobic bullying incidents.

Overall, this study contributes to our understanding of group dynamics in homophobic bullying and the role of common ingroup identities in influencing bystanders’ intentions to help. By delving deeper into the mechanisms underlying bystander responses, we can develop more targeted and effective interventions to create safer and more inclusive environments for everyone.