The effects of ethnic group membership on bullying at school: when do observers dehumanize bullies?

ABSTRACT

The purpose of this experiment was to test how ethnic group membership of both the bullies and the victims influence the way that observers attribute human characteristics to bullies. Ethnic group membership was manipulated in terms of bullies’ and victims’ ethnicity (ingroup-majority group versus outgroup-minority group). Furthermore, we examined the mediating role of empathic concern towards the victim and perspective taking of the bully in the relation between ethnic group membership of bullies and victims and attributions of humanness to the bullies. We hypothesized that observers would attribute lower human characteristics to outgroup bullies when bullies inflict harm on an ingroup victim. Moreover, we expected that perspective taking of the bully and empathic concern towards the victim would mediate this relation. Analysis of data from a sample of 293 Greek-Cypriot adolescents fully corroborated our hypotheses. The findings are discussed in terms of the discrimination-based nature of bullying at school.

KEYWORDS:

• Intergroup bullying
• ethnicity
• humanness - dehumanization
• human nature and human uniqueness
• empathy

View correction statement:

Correction

Bullying at school remains a significant problem across the world (Hinduja & Patchin, 2012; Kowalski, Giumetti, Schroeder, & Lattanner, 2014). Several studies have shown that children and adolescents involved in bullying suffer from short and long-term consequences such as internalizing and externalizing difficulties (Roland, 2002; Seals & Young, 2003).

Group processes play a significant role in bullying at school (Gini, 2006; Nesdale, 2007; Nesdale & Duffy, 2011; Nesdale, Durkin, Maass, Kiesner, & Griffiths, 2008; Palmer & Abbott, 2018). During early and middle childhood children acquire social attitudes that enable them to harm either psychologically or physically outgroup members especially when these members come from an ethnic or a racial minority (Nesdale, 2007). Bullying has been linked to low attribution of human characteristics (i.e., dehumanization) to the involved members (Fousiani, Dimitropoulou, Michaelides, & van Petegem, 2016; van Noorden, Haselager, Cillessen, & Bukowski, 2015) which justifies the enacted immoral behavior (Kelman, 1973). Importantly, racial and ethnic minorities constitute the main targets of dehumanization (Costello & Hodson, 2014; Goff, Eberhardt, Williams, & Jackson, 2008; Vaes & Paladino, 2010). Denial of humanness to others is linked to low empathy towards them, whereas empathy is considered as a requirement for overcoming dehumanization (Halpern & Weinstein, 2004). Moreover, empathy as a dispositional trait is related to low attribution of humanness to the bullies and high attribution of humanness to the victims (Fousiani et al., 2016). To the best of our knowledge, there is no research to date exploring intergroup bullying and attributions of humanness to the bullies by the observers. In this study, instead of focusing on how victims are perceived in terms of humanness by the observers, our focus lies on the effects of group membership (ingroup versus outgroup) of both the bullies and the victims in the perception of bullies as full human beings by the observers. Moreover, contrary to prior studies which have largely focused on one or the other type of humanness, namely human nature and human uniqueness (see Haslam’s model of dehumanization, 2006), we took into account both types of humanness in order to examine which one is more related with group processes of bullying. Finally, the mediating role of observers’ perspective taking of the bully and empathic concern towards the victim was explored. Interestingly, contrary to prior research, we probed the role of empathy not as a dispositional trait (Fousiani et al., 2016), but as a state (see Bauman & Del Rio, 2006) that may depend on the group membership of bullies and victims.

Bullying as an intergroup phenomenon

Bullying at school is the systematic and intentional injury or discomfort inflicted on the victim by one or more other students (Olweus, 1993). Bullying involves real or perceived imbalance of power (Rigby, 2002), with the more powerful individual or group systematically and repeatedly abusing those who are less powerful (Farrington, 1993). Bullying can be physical, verbal, or relational (Clarke & Kiselica, 1997). The view of bullying as a group process has gained relatively less empirical attention (Ojala & Nesdale, 2004). According to the social identity development theory (SIDT; Nesdale, 2007) bullying constitutes a strategy that individuals use in order to become accepted members of the ingroup and a tool to maintain and emphasize intergroup distinctions (Baumeister & Leary, 1995; Nesdale, 2007; Nesdale et al., 2008). Children learn about ingroup/outgroup relationships, they do experience threat coming from outgroups and they adjust their aggressive intentions accordingly. Specifically, Ojala and Nesdale (2004) conducted a study with student participants and found that bullying is more acceptable when directed at an outgroup member who was similar to the ingroup and therefore possibly represented a threat to the ingroup. Finally, peers show an increased focus on the ingroup and greater liking for the ingroup than the outgroup (Nesdale, 2007), and their bullying intentions are greater when the ingroup has a norm of outgroup dislike versus outgroup liking (Nesdale et al., 2008).

Surprisingly, the existing research has primarily probed intergroup bullying among social networks and friends (Nesdale & Duffy, 2011; Nesdale et al., 2008). To the best of our knowledge, how peers view ingroup and outgroup bullies or victims in terms of their race/ethnicity (same versus different ethnicity), remains at large under-explored. Race/ethnicity may be a particularly important contextual factor to examine in relation to bullying, especially given that bullying is defined as an aggressive relationship based on an imbalance of power (Olweus, 1993). In fact, research has demonstrated that peers’ bullying/victimization experiences can differ based on their race/ethnicity (Hanish & Guerra, 2000).

Bullying and recognition versus denial of humanness to the involved members

Ethnic and racial minorities, immigrants and refugees are oftentimes denied their human qualities, and thus are viewed as inferior beings (Costello & Hodson, 2014; Goff et al., 2008; Vaes & Paladino, 2010), who are not worth equal rights. In social psychology, dehumanization theory (Haslam, 2006) has been used frequently for explaining aggressive and violent behaviors against individuals or groups. Two types of human characteristics can be attributed to individuals: (a) human uniqueness (HU) characteristics (e.g., civility, morality, rationality, intelligence) distinguish humans from animals and involve high order cognition. Denial of HU traits to individuals is called animalistic dehumanization (Haslam, 2006) and it involves categorization of others as inferior beings, hence justifying aggressive or violent behaviors against others; (b) human nature (HN) characteristics (e.g., emotional responsiveness, cognitive openness) distinguish people from machines or automata. Denial of these characteristics to others is called mechanistic dehumanization (Haslam, 2006) and it involves viewing the others as emotionally cold, close-minded and passive just like objects and it serves in treating them with psychological distance or indifference.

There is limited research evidence linking anti-social behavior and bullying at school with such dehumanizing practices (Costello & Hodson, 2014; Fousiani et al., 2016; van Noorden et al., 2015; van Noorden, Cillessen, Haselager, Lansu, & Bukowski, 2017). Van Noorden and colleagues (2015) found that denial of HU to peers is more common than denial of HN and that non friends are attributed lower HU than friends. Importantly, denial of both HU and HN traits to peers was associated with victimization of dehumanized peers. Moreover, bullies, victims and bully/victims attribute fewer human characteristics to each other than noninvolved members (van Noorden, Bukowski, Haselager, Lansu, & Cillessen, 2016). Apart from attribution of humanness characteristics to bullies or victims by an observer, prior research has also investigated the actual possession of humanness traits by bullies. According to the literature, bullies seem to score lower in human uniqueness characteristics but not in human nature traits. Low HU of people fosters moral disengagement and delegitimization of victims and can therefore explain why individuals scoring low in HU engage in bullying behaviors (Fousiani et al., 2016). To the best of our knowledge there is no study to date investigating these effects in an ethnicity-based intergroup context.

In this study we hypothesized a differential attribution/denial of humanness to bullies by the observers, depending on the group membership (ingroup versus outgroup ethnicity) of both the bullies and the victims. Despite the focus of the literature on the importance of HU over HN traits in bullying phenomena (Fousiani et al., 2016; van Noorden et al., 2015), we claim that denial of both senses of humanness can occur in intergroup bullying. Both types of humanness are “…complementary dimensions of social judgment that contribute in specific ways to the way we perceive and behave towards others” (Vaes, Leyens, Paladino, & Miranda, 2012, p. 42). Groups use one or the other dimension to distinguish themselves according to the specific context of comparison (Bain, Park, Kwok, & Haslam, 2009; Bain, Vaes, Haslam, Kashima, & Guan, 2012). Moreover, recent research has demonstrated the preponderance of HN traits in situations of suffering and victimization (Sakalaki, Richardson, & Fousiani, 2017). HN involves characteristics such as emotional responsiveness, warmth, cognitive openness, and depth (Haslam, 2006) that are essential in intergroup relations (Saguy et al., 2015). The observer would deny to attribute these traits to members involved in aggressive behaviors, and more so when aggressive behaviors are directed towards members they identify with. In line with this reasoning, we expected that observers would attribute not only low HU but also low HN traits to bullies depending on the group membership of both bullies and victims.

Based on the SIDT (Nesdale, 2007) and the self-categorization theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987) we hypothesized that observers would attribute lower HU but also lower HN traits to outgroup as compared to ingroup bullies (Hypothesis 1a). Importantly, this would mostly be the case when outgroup bullies inflicted harm on an ingroup rather than an outgroup victim (Hypothesis 1b). On the contrary, higher attribution of humanness (both UH and HN) should be observed when the bully is an ingroup than an outgroup member, and especially when the victim comes from an outgroup than the ingroup. We did not state specific hypotheses regarding attribution of humanness to the bully when both bullies and victims come from the same group.

The mediating role of empathy

Recognizing the others as full human beings is a prerequisite for perceiving and being moved by the others’ pain and suffering; instead denial of humanness to others is linked to decreased empathy towards others (Halpern & Weinstein, 2004). As far as the role of empathy in victimization situations is concerned, there is evidence that when people do recognize human characteristics to victims, they experience more empathy towards them (Capozza et al., 2013; Čehajić, Brown, & Gonza´Lez, 2009). Interestingly, empathy as a personality trait serves as a buffer against bullying behavior (Kokkinos & Kipritsi, 2012). According to the definition of empathy by Gini (2006), two main dimensions of empathy can be distinguished: (a) an affective dimension, often referred to as empathic concern, which represents the ability to experience another’s emotions; and (b) a cognitive dimension, often referred to as perspective taking, which reflects the ability to understand another person’s emotional state (Gini, 2006). Empathic concern as a personality variable, has been related to lower recognition of humanness to the bullies and higher recognition of humanness to the victims (Fousiani et al., 2016). In the current study, instead of measuring empathy as a personality trait, we focused on adolescents’ empathic concern towards the victim and on their perspective taking towards the bully as states that might depend on the group membership of both the victim and the bully. We hypothesized that the effect of ethnic group membership of both the bully and the victim on attribution of humanness to bullies would be mediated by a) the empathic concern towards the victim and b) the perspective taking of the bully (see Figure 1). More specifically, observers should display higher empathic concern towards an ingroup than an outgroup victim, while this should be mostly the case when the victim has been harmed by an outgroup rather than an ingroup bully. Higher empathic concern towards the victim would in turn, lead the observers attribute lower HU and HN to the outgroup bully (Hypothesis 2). Additionally, observers would display lower perspective taking towards an outgroup rather than ingroup bully; especially when the bully has harmed an ingroup than outgroup victim. The observer would in turn, attribute lower HU and HN to the outgroup bully (Hypothesis 3).

Figure 1. Diagram of the mediation model.

Importantly, our study was based on an experimental design where participants read several vignettes presenting the main types of bullying, namely physical, verbal, and relational (Bauman & Del Rio, 2006). Group membership of both the bullies and the victims was operationalized in each of these vignettes. This study took place in the Republic of Cyprus, a multicultural country where several minorities co-exist together with the majority group, the Greek-Cypriots. Most of our participants were Greek-Cypriot school students so in order to operationalize group membership of bullies and victims, ingroup members (either bullies or victims) were always presented as having a Greek-Cypriot nationality. On the other hand, Russian-Pontians constituted the outgroup members (either bullies or victims). Russian-Pontians, although they have Greek ancestry and originally are called Pontian-Greeks, are discriminated in Cyprus and are seen as a different ethnicity because of their migration history to other parts of Eastern Anatolia, to the former Russian province of Kars Oblast in the Transcaucasus, and to Georgia, in various waves between the Ottoman conquest of the Empire of Trebizond in 1461 and the second Russo-Turkish War of 1828–1829 (Wood, 2005). For this reason, instead of Pontian-Greeks they are known as “Russians-Pontians”, which has a discriminative connotation. Russian-Pontians’ mother language is Russian while Russian-Pontians are faced with social prejudice (Anti-discrimination - social inclusion of Pontian Greeks in the town, January, 2010).

Method

The sample consisted of 335 adolescent students in grades 10 or 11 from 11 different Lyceums (upper high schools) in Nicosia, the capital of Cyprus. We excluded 2 students did not report their grade, and 40 students who did not report Greek-Cypriot (the ingroup in the vignettes that were used in the experimental manipulations) as their nationality. The final sample consisted of 293 students (55.6% female, 47.4% 10^th graders, 81.2% from married families). All participants lived in Cyprus on a permanent basis and spoke Greek.

Experimental procedure

For conducting the study, the first step included a permission obtained by the Cyprus Ministry of Education and Culture as well as the Pedagogical Institute of Cyprus, which is responsible for research affairs in schools in Cyprus. Having ensured that the procedure follows the prerequisite criteria and the regulations about ethical issues, questionnaires were administered and filled out during a regular class period at schools. The research was based on purposive sampling, since the number of classroom grades were selected depending on the availability of each school unit. Participation in the study was voluntary and anonymity and confidentiality were guaranteed.

A questionnaire including vignettes and scales was constructed and administered to respondents. All materials were translated into Greek following the procedure of back translation. For the present study the questionnaire was comprised of the following parts ¹ :

Experimental design and materials

Ethnic group membership (ingroup/outgroup) of bullies and victims was manipulated in a between-subjects design. The study included six vignettes following the paradigm of Bauman and Del Rio (2006). The vignettes referred to three different types of bullying, verbal, relational, and physical. Two of the vignettes portrayed an incident of physical bullying, two described a verbal bullying event, and the remaining two presented social exclusion which is a type of relational bullying (Yoon & Kerber, 2003). All six vignettes were presented to the participants in random order. All vignettes were adjusted to the specifics of the current study. Specifically, in each vignette we included information about the bully’s and the victim’s ethnic group membership (ingroup/outgroup member). For the manipulation of the bully’s and victim’s ethnicity we created three between-group experimental conditions in total, namely, Ingroup victim – ingroup bully (N = 96, 32,8%), Ingroup victim – outgroup bully (N = 95, 32,4%), Outgroup victim – ingroup bully (N = 102, 34,8%). Some sample vignettes were as follows:

[Verbal Bullying] “At the writing center you hear a Russian-Pontian student chant to a Greek-Cypriot child, “Teacher’s spit, brown-nose, suck-up, kiss-ass.” The child tries to ignore the remarks but sulks at his desk. You saw this same thing happen the other day…”

[Relational bullying] “During project time you overhear a Greek-Cypriot child say to another Greek-Cypriot student, “If you don’t let me have the purple marker, I won’t invite you to my birthday party.” This is not the first time you have heard this child say this type of thing…”

[Physical bullying] “A Greek-Cypriot student brought a dinosaur shaped eraser to school. He boasts that it was a prize from a game arcade. A Russian-Pontian child goes over and smacks his head, demanding the eraser. The child refuses at first, but eventually gives in…”

In the six vignettes that each participant read, the ethnicities of the bully and the victim were kept constant for each participant, but differed between the three conditions. After reading each single vignette participants were asked to fill in 1 item measuring empathic concern towards the victim and 1 item measuring perspective taking of the offender. Consequently, respondents filled in the empathic concern as well as the perspective taking item that followed each vignette six times in total. A mean was calculated for the empathic concern items as well as for the perspective taking items. At the end, and after having read all six vignettes, participants were asked to fill in a scale measuring attribution of (HN) traits to the bullies and a scale measuring attribution of (HU) to the bullies. The items included in both scales were adjusted to the ethnicity of the bullies depending on the experimental condition. At the end, participants indicated their demographic characteristics and finally they were thanked for their participation and debriefed.

Measures

Empathic concern and perspective taking

For measuring empathetic concern (EC), the affective dimension of empathy towards the victims, each vignette was followed by the question (six items in total for all the vignettes): “I would be upset by the [bully’s ethnicity] student’s behavior and feel sympathetic to the [victim’s ethnicity] child”. EC towards the victims was the only measure we used in this study that refers to the victims while the rest of the scales refer to the bullies. The cognitive dimension of empathy, that is perspective taking (PT) of the bully was assessed with one item following each vignette (six in total) which was: “I would try to be in the [bully’s ethnicity] student’s position and understand the reasons for which he/she behaved this way towards his [victim’s ethnicity] classmate”. Both items were rated on a 7-point Likert scale ranging from 1 (“strongly disagree”) to 7 (“strongly agree”) and Cronbach’s was .86 and .89 for each set of questions respectively. Both questions have been used by Bauman and Del Rio (2006) in a previous study.

Attribution of human traits to the bullies

Participants were asked to rate the extent to which they attributed both high HN and high HU characteristics to the bullies. The scales were based on measures already used in prior studies (Bastian & Haslam, 2010; Fousiani et al., 2016; Haslam, 2006). High HN characteristics were described in 3 items such as the following examples “the Greek-Cypriot/Russian-Pontian student… is open minded, like he could think clearly”, or that the Greek-Cypriot/Russian-Pontian student is emotional, responsive and warm in his interpersonal relations”. As far as high HU characteristics are concerned 4 items were used characterizing, for example, bullies as “rational and logical”, or that they “have self-restraint”. Responses were given from 1 ‘‘strongly disagree’’ to 7 ‘‘strongly agree’’. Cronbach’s alpha were .85 and .74 for bullies’ UH and HN characteristics scales respectively. Both HU and HN scales have been already used in a Greek-speaking sample in the past (Fousiani et al., 2016).

Demographic characteristics

The last part of the questionnaire involved gathering information related to demographic variables of the participants including their sex, grade level, their nationality (Cypriots or non-Cypriots), and country of birth.

Analysis procedure

Following descriptive statistics of the main study variables, a MANOVA test was conducted to investigate the effects of the independent variable (group membership of bullies and victims) on the two dependent variables HN and HU. For any significant differences between levels of the independent variable, a conditional process analysis was employed (Hayes, 2013) to study the possibility of a mediation effect of PT and EC (Figure 1). Mediation analysis was conducted using 5000 bias-corrected bootstrap confidence intervals in PROCESS 2.16 for SPSS (Hayes, 2013).

Results

Descriptive statistics for the main variables of the study appear on Table 1. On the 7-point response scale that was used for all variables, the participants endorsed on average low ratings for HN, HU and PT for the bullies in the vignettes, and high ratings of EC for the victims. The lowest average ratings for HN, HU and PT for the bullies and the highest average rating for EC for the victims were found in the Outgroup bully-Ingroup victim condition. The highest average ratings for HN and HU for the bullies and the lowest average for EC for the victims were found in the Ingroup bully-Outgroup victim condition. The highest PT for the bullies was reported in the Ingroup bully-Ingroup victim condition. The HN and HU measures were strongly correlated (r = .76) and both correlated moderately with PT (r = .32 for HN and .29 for HU). EC for the victim was negatively associated with all other variables: r = -.35 with HN, -.42 with HU, -.22 with PT (p < .001 for all r’s).

Table 1. Means (and SD) for the study variables by experimental condition and for the overall sample.

Condition	N	Variables
		Human Nature (B)	Human Uniqueness (B)	Perspective Taking (B)	Empathic Concern (V)
Ingroup bully-ingroup victim	96	2.49 (1.57)	2.42 (1.40)	2.89 (1.65)	5.78 (1.08)
Outgroup bully-ingroup victim	95	2.16 (1.46)	2.06 (1.31)	2.24 (1.35)	6.29 (0.85)
Ingroup bully-outgroup victim	102	2.95 (1.70)	2.72 (1.68)	2.69 (1.70)	5.57 (1.46)
Overall sample	293	2.55 (1.61)	2.41 (1.50)	2.61 (1.60)	5.87 (1.20)

Note. B: for bullies, V: for victims.

A MANOVA test with ingroup/outgroup group membership as an independent variable and the HN and HU as dependent variables was conducted. Box’s test for the equality of covariances was not significant (Box’s M = 10.146, p = .123), while the Levene’s tests for the two outcomes were significant (F_{2, 290} = 3.779, p = .024 for HN; F_{2, 290} = 4.015, p = .019). The two dependent variables did not follow normal distributions in each group, but the large samples in the three conditions ensured a robust MANOVA (cf. Tabachnick & Fidell, 2007). The multivariate effect was significant, Pillai’s Trace = .044, p = .011, η_p ² = .022 indicating a significant difference on the combination of the outcomes. Separate univariate effects were significant for HN, F_{2, 290} = 6.173, p = .002, η_p ² = .041, and for HU, F_{2, 290} = 4.860, p = .008, η_p ² = .032. Due to the violation of the equal variances assumption, Games-Howell post hoc tests were carried out. Two significant differences were found: The Ingroup bully-Outgroup victim compared to the Outgroup bully-Ingroup victim conditions resulted in higher mean differences on HN (mean difference = .79, SE = .23, p = .002) and on HU (mean difference = .66, SE = .21, p = .007). In the Ingroup bully-Outgroup victim condition, observers attributed significantly higher human characteristics to the bully compared to the Outgroup bully-Ingroup victim condition where those attributions were lower. No significant differences were found between the Ingroup bully-Ingroup victim and the other two conditions.

For the two conditions that differed significantly on the two outcomes, separate mediation models were conducted: ingroup/outgroup membership was the independent variable (0: Outgroup bully-Ingroup victim, 1: Ingroup bully-Outgroup victim), HN or HU were the dependent variables, and PT for the bully or EC for the victim were specified as mediators. The total effect of ingroup/outgroup group membership on HN was positive and significant: the bully was more humanized in the Ingroup bully-Outgroup victim, or was dehumanized in the Outgroup bully-Ingroup victim condition (Table 2). When PT for the bully was added as a mediator in the model, both the direct and the indirect effects were significant, resulting in partial mediation. A small part of the effect of group membership of the bully and victim on HN was mediated by how much the observer could take the perspective of the bully as indicated by the ratio of indirect to total effect. Very similar results were found when HU was the dependent variable and PT was the mediator. Looking at the unstandardized path coefficients, higher PT of the bully was found at the Ingroup bully-Outgroup victim condition compared to the Outgroup bully-Ingroup victim condition, and higher PT predicted higher attributions of humanness.

Table 2. Mediation results with PT and EC as mediators.

			Unstandardized paths		Indirect effect
Effects of condition on	Total effect	Direct effect (c΄)	a	b	Estimate	BCA CI	Ratio of indirect to total effect
HN (mediator PT)	.79 (.23)***	.66 (.22)**	.45 (.22)*	.30 (.07)***	.13 (.07)	.02-.29	.17
HN (mediator EC)	.79 (.23)***	.46 (.22)*	−.72 (.17)***	−.46 (.09)***	.33 (.10)	.16-.57	.42
HU (mediator PT)	.66 (.22)**	.54 (.21)*	.45 (.22) *	.26 (.07)***	.12 (.06)	.02-.25	.18
HU (mediator EC)	.66 (.22)**	.30 (.21)	−.72 (.17)***	−.49 (.08)***	.35 (.10)	.19-.59	.54

Notes. Standard errors in parentheses (bootstrap standard errors for the indirect effect estimate); BCA CI: bias-corrected and accelerated bootstrap confidence interval; paths a and b correspond to the prediction coefficients of the independent variable to the mediator (path a) and of the mediator to the dependent variable (path b), see Figure 1; * p < .05, ** p < .01, *** p < .001.

When EC for the victim was added as a mediator, the total effect on both HN and HU was positive and significant. In the case of HN, partial mediation was found. The group membership of the bully and victim had both a direct and an indirect effect on HN via EC. In this case the ratio of the indirect to the total effect was larger than before. When HU was the outcome, a full mediation was found. The effect of group membership of the bully and victim on HU was fully mediated by EC, indicating that the affective component of empathy plays a more important role in attributions of humanness when the group membership of bully and victim is manipulated. As can be seen through the unstandardized path coefficients, lower EC towards the victim was found at the Ingroup bully-Outgroup victim condition compared to the Outgroup bully-Ingroup victim condition, and EC for the victim negatively predicted attributions of humanness for the bully.

Discussion

This study aimed to explore how observers make attributions of human characteristics to the bullies (versus dehumanize bullies) depending on the group membership, in terms of ethnicity, of both the bullies and the victims. Importantly, we took into account both types of humanness as presented in the model of dehumanization of Haslam (2006), namely human nature and human uniqueness. It is noteworthy that this study was an experimental design where all participants read several scenarios presenting physical, verbal, and relational bullying. The study was conducted in Cyprus, where intergroup relations issues between different ethnicities are prevalent. Cyprus, except of being a divided country since 1974 (Greek-Cypriots live in the southern area and Turkish-Cypriots in the northern area of Cyprus), is also a multicultural country where several minorities, such as Armenians, Maronites, and Pontians among others, live together with the majority group (Greek-Cypriots) and are quite marginalized (Anti-discrimination - social inclusion of Pontian-Greeks in the town, January, 2010).

The findings showed that observers attribute lower HU but also HN to outgroup than ingroup bullies, especially when the victims come from the ingroup rather than the outgroup. These results corroborate our hypotheses 1 and 2 and are in line with the claims of the social identity development theory (Nesdale, 2007) and the social categorization theory (Turner et al., 1987). Although prior research has investigated bullying in intergroup contexts, the existing literature primarily focuses on intergroup bullying among social networks and friends (Nesdale & Duffy, 2011; Nesdale et al., 2008), neglecting the role of ethnicity. To the best of our knowledge, this is the first study to experimentally test group processes of bullying at school considering the role of ethnicity.

Moreover, prior studies have pointed out the preponderance of HU over HN characteristics in bullying situations (Fousiani et al., 2016; van Noorden et al., 2015). This is the first study to show that both humanness dimensions, that is HU and HN, play a significant role in bullying and victimization phenomena. HU, on the one hand, involves traits such as morality, intelligence, civility and refinement. These traits may be denied to bullies and bullies may be seen as inferior persons equal to animals (animalistically dehumanized). HN, on the other hand, involves characteristics such as warmth, responsiveness, depth and agency, which may be also denied to bullies and thus bullies may be seen as unemotional and superficial beings equal to machines or automata (mechanistically dehumanized). Our findings are in line with prior literature claiming that both types of humanness are complementary dimensions of social judgment (Vaes et al., 2012, p. 42) and groups use one or the other dimension to distinguish themselves (Bain et al., 2009; Bain, Vaes, Kashima, Haslam, & Guan, 2012).

Interestingly, the mediating role of empathy in the relation between group membership of bullies and victims and attributions of humanness to the bullies was also investigated. We took into account both the empathic concern and perspective taking, so that we include both the affective and cognitive dimensions of empathy (Gini, 2006). Most importantly, as far as victims are concerned, we focused on the empathic concern towards the victims since literature has revealed the significance of affective over cognitive empathy towards victims in bullying contexts (Stavrinides, Georgiou, & Theofanous, 2010). However, we also considered the perspective taking of bullies as observers would more easily try to take the perspective of a harm-doer than emotionally identify with them. The findings supported our second hypothesis showing that observers display higher empathic concern towards an ingroup victim who has been harmed by an outgroup bully and thus attributes lower humanness to the bully. Similarly, we found that observers display lower perspective taking towards an outgoup bully who has harmed an ingroup victim, attributing in turn lower humanness to the bully. These findings confirmed our third hypothesis.

The findings reveal that bullying at school may have roots in discrimination and prejudice. These findings are particularly important because they are obtained in Cyprus, a divided country since the occupation of part of the country by Turks in 1974. Cyprus faces intergroup conflicts between Greek-Cypriots (i.e., the majority group) and Turks and Turkish-Cypriots (i.e., minority group) and has seen significant changes in its demographic composition in the last twenty years (from the Philippines, Georgia, Russia, Romania, Sri Lanka among others). Several attempts have been made in order to diminish aggression and violence in Cyprus and foster intergroup reconciliation and peace (Psaltis, 2012). The “Home for Cooperation” in Cyprus (http://www.home4cooperation.info/what-is-the-h4c) is a community center for the promotion of intercommunal cooperation and intercultural dialogue through the development of anti-discrimination programs. Our findings might be of interest to such institutions not only to fight intergroup violence in general but also violence in adolescents in particular. Moreover, new intervention programs against bullying at school in particular, have been recently implemented in Cyprus. For instance, the “ViSC” program, a socio-ecological anti-bullying program that has been developed, implemented and rigorously evaluated in Austria, has been recently implemented in Cyprus (Solomontos-Kountouri, Gradinger, Yanagida, & Strohmeier, 2016). Our findings could contribute to the enrichment of this program so that intergroup phenomena in bullying are better understood.

Limitations and future directions

This study has a number of important limitations that should be noted. First, the use of self-report instruments for the assessment of both humanness dimensions and empathy may have led to an over-estimation of the association between group membership of bullies and victims and the mediating and outcome variables. Future studies should develop and pilot test scales measuring attributions of humanness in adolescents population. Further, the data were collected by a particular age period as well as in a specific cultural context, which may limit the generalizability of the findings. Cyprus is a relatively collectivisticly oriented country, where the expression of aggression and violence may be displayed differently as compared to the individualistic ones. Similarly, it could be interesting to test whether the findings also generalize to more clinical populations. Future research could explore these effects in individualistic contexts as well. Additionally, in our experimental design we did not include a condition outgroup victim/outgroup bully, which eventually created some constraints in the effects we could test. Future studies should replicate these results using the full experimental design. Finally, researchers should investigate the effect of group membership on humanness attributed to the victims instead of bullies. The way victims are viewed by either the bullies or the observers depending on their national background has been largely neglected and should provide important information with regards to intergroup bullying phenomena.

Conclusion

Taken together, the findings obtained in this study shed light on the important role of ethnic group membership, in (de)legitimizing bullies by ascribing to them full humanness. Results showed that observers are biased in the way they perceive harm-doers depending on the ethnical background of the persons involved in the harm-doing. Bullying, when enacted by ingroup bullies seems to be more acceptable, especially when the targets are outgroup victims; observers can more easily take their perspective, possibly understand and rationalize their motives, and attribute to them higher humanness as compared to outgroup bullies who harm ingroup victims. These findings point out that ethnicity based biases lead to unfair treatment of adolescents from minority ethnic backgrounds.

Conflict of Interest

Authors declare that they have no conflict of interest.’

Compliance with Ethical Standards

This research involves human participants. All procedures performed in this study were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Data Availability Statement

The data described in this article are openly available in the Open Science Framework at https://doi.org/10.17605/OSF.IO/YVE8F

Open Scholarship

This article has earned the Center for Open science badges for Open Data and Open Materials through Open Practices Disclosure. The data and materials are openly accessible at https://doi.org/10.17605/OSF.IO/YVE8F

Notes

^1. Sufficient information to reproduce the reported methodology is available from the Open Science Framework at https://doi.org/10.17605/OSF.IO/YVE8F.