Cyberbullying, defined as repeated online behavior that is intended to cause discomfort or harm to others (e.g., threats, name calling, exclusion, rumor spreading) (Smith, 2016), has become a widespread social problem among adolescents and adults (Kowalski et al., 2019). Encouraging bystanders (i.e., witnesses) to intervene (i.e., bystander defending) has been identified as a critical strategy to reduce cyberbullying and mitigate negative outcomes (Myers & Cowie, 2019; Torgal et al., 2021). Whereas previous defending research has mainly focused on adolescents (for a review, see Lambe et al., 2019), cyberbullying is a concern for students at each developmental stage, from school to university (Myers & Cowie, 2019). Indeed, research indicates that 36 to 69% of university students witness cyberbullying, while the majority (61%) remain passive bystanders, taking no action to intervene (Gahagan et al., 2016). In addition, bystanders’ empathy for victimized peers has been found to decrease with age, with university students showing the least sensitivity to peer’s distress (Myers & Cowie, 2019). Research is therefore needed to identify how to promote appropriate and harm-reducing responses among university bystanders.

Understanding how to promote appropriate prosocial bystander responses is also important as recent studies highlight that bystander defending behavior in response to both in-person and cyberbullying is multifaceted, including both aggressive and prosocial defending strategies (e.g., Bussey et al., 2020; Lambe & Craig, 2020). Although both forms of defending may be motivated by a desire to stop bullying and help the person who is victimized, aggressive defending involves antisocial strategies such as confrontation, using threats or saying mean things about the person doing the bullying (Meter et al., 2019; Salmivalli et al., 1996). In contrast, prosocial defending involves constructive strategies such as comforting the person being victimized or reporting the bullying, which facilitate peaceful conflict resolution. The overarching aim of the present study was to better understand empathic and compassionate responses associated with different bystander behavioral intentions in response to cyberbullying, namely passive bystanding, aggressive defending, and prosocial defending.

Empathic Responses and Bystander Behaviors

Empathy, defined as the ability to understand (i.e., cognitive empathy) and vicariously share (i.e., affective empathy) the emotional states of others, is considered critical for prosocial behavior (Decety et al., 2016). Research has shown that empathy in response to bullying is positively associated with defending (for a systematic review, see Lambe et al., 2019), and many bullying interventions include empathy training (Cefai et al., 2018; Gaffney et al., 2019). It is surprising, however, that most of these interventions do not reflect the possibility that empathy can trigger individual differences in internal emotional and cognitive responses when witnessing bullying, which are also expected to be important to bystander responses. For instance, research has shown that empathy for another’s distress can facilitate empathic distress, empathic anger, and/or compassion, which in turn relate to different interpersonal behaviors (e.g., Singer & Klimecki, 2014; Stevens & Taber, 2021). Here, we investigated the associations of empathic distress, empathic anger, and compassion with different bystander behavioral intentions in response to witnessing cyberbullying in order to encourage the refinement of empathy-related intervention content and enhance intervention effectiveness.

Empathic Distress

Excessive sharing of another’s distress has been described as empathic distress (or personal distress) (Batson et al., 1987; Singer & Klimecki, 2014). Empathic distress reflects an involuntary, self-oriented emotional response characterized by discomfort and uneasiness when in the presence of another’s distress, which is argued will prevent the ability to differentiate own feelings from those witnessed in others. Indeed, research has shown that feelings of empathic distress are associated with egoistic or self-oriented motivation to avoid another’s suffering to minimize or protect oneself from aversive feelings (Eisenberg & Fabes, 1990; Klimecki, 2019). This is not surprising given that feelings of distress have evolved to trigger fight-flight-freeze responses to increase the chances of survival in circumstances that threaten life or social safety (Mobbs et al., 2015).

This framework is also consistent with the empathy-altruism hypothesis (Batson et al., 1983, 2015), which states that empathic distress motivates helping only when it aids in reducing personal negative emotions in situations where escape is not possible. In line with this view, research examining a variety of prosocial helping behaviors has shown that empathic distress in response to witnessing other’s suffering (e.g., injured children) is related to a greater chance of avoidance and only positively associated with helping if it is difficult to escape the situation (e.g., Eisenberg et al., 1989; Hein et al., 2010). Accordingly, empathic distress may only motivate helping associated with low costs or in situations that are difficult to avoid.

There can be costs associated with defending a victim of bullying, such as being judged by others and becoming a target of victimization (see, e.g., Padilla-Walker et al., 2018; Spadafora et al., 2020). Thus, passive bystanding may occur when costs are deemed to be high and witnesses need to regulate uncomfortable feelings (e.g., feelings of helplessness and worry). Yet, the research on empathic distress and bystander responses to bullying remains limited. The current study is the first to examine whether empathic distress is associated with passive bystanding intentions and different forms of defending intentions in response to witnessing cyberbullying, while also controlling for empathic anger and compassion. We expected a positive association between empathic distress and passive bystanding only.

Empathic Anger

Consistent with Hoffman’s theory of empathy (2001), when witnessing the unfair and/or intentionally harmful treatment of others, empathy may take the form of anger on behalf the person being victimized (i.e., other-condemning moral emotion referred to as empathic anger or third-party anger). According to Hoffman, empathic anger activates one’s motivation to defend the person being victimized and decreases the likelihood of passive bystanding. Extending this view, Vitaglione and Barnett (2003) were the first to show that self-reported empathic anger in adults is not only associated with a desire to help individuals who have been intentionally hurt, but also a desire to punish the person who offends. This finding has also been supported in more recent bystander helping studies (e.g., Gummerum et al., 2020; van Doorn et al., 2018).

Given that bullying involves the intentional victimization of others, empathic anger may serve as an important motivator of moral action that evokes both prosocial and aggressive defending and thus also inhibits passive bystanding. Consistent with this view, Pozzoli et al. (2017) demonstrated that students’ (mean age = 12 years) self-reported empathic anger was negatively associated with their self-reported passive bystanding and positively associated with self-reported general defending. In addition, a recent study found that adolescents’ vicarious anger in response to witnessing social exclusion in a cyberball virtual reality game was positively associated with their prosocial defending (i.e., comforting) enacted in the game (Lambe & Craig, 2022). The current study expands on this research by examining whether these associations are replicated for university students in a cyberbullying context, expecting that empathic anger will be associated with lower passive bystanding intentions and greater defending intentions (both aggressive and prosocial).

Compassion

When witnessing the suffering of others, bystanders may also respond with compassion. Compassion is a complex construct, and there is not one universally agreed-upon definition. Consistent with recent research and conceptualizations, compassion is characterized by distress tolerance and other-oriented concern for all people, believed to facilitate prosocial action to help suffering other (for reviews see Gilbert, 2020; Klimecki, 2019; Stevens & Taber, 2021). Accordingly, it is believed to involve the ability to notice and remain open to one’s own and others’ distress without suppression or overidentification, typically referred to as mindfulness or distress tolerance (e.g., Gilbert et al., 2017; Gu et al., 2017). Hence, compassion implies acceptance and tolerance of difficult empathic emotions, such as distress or anger, which may mitigate against fight-flee-freeze responses when witnessing another’s distress or the unfair treatment of a person being victimized (Gu et al., 2017; Stevens & Taber, 2021).

A compassionate response should therefore facilitate a focus on the person being victimized rather than on personal aversive feelings. Such an other-oriented attentional focus is believed to facilitate feelings of concern for the person being victimized and more constructive helping (Klimecki, 2019; Singer & Klimecki, 2014). This view aligns well with Batson et al.’s empathy-altruism hypothesis (1983, 2015), which states that, in contrast to self-oriented feelings of empathic distress, other-oriented feelings of concern motivate altruistic helping (i.e., helping regardless of costs). This is consistent with the view of compassion as accompanied by feelings of warmth and concern for others’ wellbeing (i.e., empathic concern), as well as an openness and non-judgmental attitude toward their suffering and inadequacies regardless of differences (Gu et al., 2017).

Witnesses of bullying may respond with compassion, but this has received little research attention. More generally, research has shown that higher levels of self-reported compassion for others, as well as the experimental induction of compassion (loving-kindness meditation) are positively related to generosity, cooperation, and altruistic helping (e.g., Weng et al., 2015), a concern for all people, even disliked others (Oveis et al., 2010; Sprecher & Fehr, 2005), and a lower likelihood of punishing individuals who offend others (McCall et al., 2014). In the context of bullying then, compassion may be positively associated with prosocial defending and negatively associated with aggressive defending and passive bystanding, especially after simultaneously considering empathic distress and anger. However, no study to date has examined the associations of compassion (as characterized by distress tolerance and other-oriented concern for all people) with bystander responses to bullying.

The Present Study

In the present study, we investigated the associations of empathic distress, empathic anger, and compassion with the three bystander behavioral intentions of passive bystanding, aggressive defending, and prosocial defending. While previous research on defending has mainly focused on adolescents (for a review, see Lambe et al., 2019), this study included a sample of university students, acknowledging that cyberbullying remains a significant problem into university (Myers & Cowie, 2019; Stevens & Taber, 2021). As previous research has shown that defending is associated with personal experiences of victimization and bullying (e.g., Bussey et al., 2020; Lambe et al., 2019), we controlled for participants’ history of these in the multivariate analyses. Further, because of gender differences in previous bullying research, with females showing more empathy and defending than males (Lambe et al., 2019), we included gender as an additional covariate. Finally, given that the self-report nature of this study and the use of socially sensitive items (i.e., measures of empathic responses and bystander behavioral intentions) may contribute to socially desirable responses, social desirability was included as an additional covariate to improve the validity of our measures. All associations between the study variables were explored, and three specific hypotheses were tested:

  • H1. (a) Empathic distress will be positively associated, and (b) empathic anger and (c) compassion will be negatively associated, with passive bystanding intentions.

  • H2. (a) Empathic anger will be positively associated, and (b) compassion will be negatively associated, with aggressive defending intentions.

  • H3. (a) Empathic anger and (b) compassion will be positively associated, with prosocial defending intentions.

Method

Participants

Study participants were 270 university students, enrolled in a first-year psychology course. Participants were aged between 18 and 29 years (Mage = 20.34, SD = 2.78, 74.8% females), and the ethnic composition of the sample was 71.9% Australian/New Zealander/European, 3.7% Australian First Peoples/Torres Strait Islander/Pacific Islander, 14.8% Asian, and 2.2% African, with the remaining from other ethnic groups. All participants were members of at least one social networking site (Facebook = 94.7%, Instagram = 90.6%, Twitter = 18.8%, Snapchat = 86.8%). Another five participants attempted the survey, but one participant was excluded from the analyses for failing two attention check items, and four participants were excluded because of excessive missing data.

Measures

Empathic Distress and Anger

Empathic distress was measured with the 7-item Personal Distress subscale of the Interpersonal Reactivity Index (Davis, 1983; e.g., “when I see someone gets hurt, I tend to remain calm” [reversed]). Participants indicated how well each item describes them on a scale ranging from 0 (not at all) to 4 (extremely). An empathic distress score was calculated by reversing some items and then averaging all items, with a higher score indicating a higher level of empathic distress; Cronbach’s α = .78.

Empathic anger was measured using the 7-item Trait Empathic Anger scale (Vitaglione & Barnett, 2003; e.g., “I feel angry for a person when his or her feelings have been hurt by someone else”). Participants indicated their agreement with each item on a scale ranging from 0 (does not describe me very well) to 4 (describes me very well). A total empathic anger score was calculated by reversing necessary items and then averaging the items, with a higher score indicating a higher level of empathic anger; Cronbach α = .87.

Compassion

Compassion was measured using the 10-item Compassion for Others subscale of the Compassionate Engagement and Action Scale (Gilbert et al., 2017). This subscale taps the ability to be compassionate to distressed others and the compassionate attributes of sensitivity to suffering, non-judgment, distress tolerance, concern for others’ wellbeing, and commitment to engage with others’ suffering (e.g., “I am accepting, non-critical, and non-judgmental of other people’s distress”). Three reverse filler items were also included to prevent distortion of responses and response bias (Gilbert et al., 2017). Participants rated each statement according to how frequently it occurred on a scale from 1 (never) to 10 (always). The 10 compassion items were averaged to create a composite compassion score, with a higher score indicating more compassion; Cronbach’s α = .89.

Cyberbullying Bystander Behavioral Intentions

Bystander behavioral intentions in response to witnessing cyberbullying were measured with 22 items adapted from previous measures (i.e., Bastiaensens et al., 2014; Salmivalli & Voeten, 2004). Adaptations involved selecting items that represented passive bystanding, aggressive defending, and prosocial defending. After being presented with a definition of cyberbullying on social networking sites,Footnote 1 participants indicated, on a scale from 1 (not likely at all) to 5 (very likely), how likely they would be to engage in each behavior if they witnessed someone being bullied on a social networking site (7 passive bystanding items, e.g., “ignore the situation”; 4 aggressive defending items, e.g., “attack the bully in order to defend the victim”; and 10 prosocial defending items, e.g., “try to comfort the victim”). A principal component analysis with an oblique rotation supported three factors of passive bystanding, aggressive defending, and prosocial defending. The factors explained a total of 55% of the item variance. Item loadings and eigenvalues are shown in Table 1. Total scores were calculated by averaging items that loaded highly on each factor, with higher scores indicating a greater likelihood of engaging in the relevant bystander behavior: Cronbach α = .88 for passive bystanding, .72 for aggressive defending, and .89 for prosocial defending.

Table 1 Item loadings and eigenvalues for passive bystanding, aggressive defending, and prosocial defending (N = 270)
Full size table

Cyberbullying and Victimization History

An adapted version (see Gámez-Guadix et al., 2014) of the Cyberbullying Questionnaire (CBQ; Calvete et al., 2010) was used to measure personal experiences of cybervictimization and cyberbullying. Participants indicated, on a scale from 0 (never) to 4 (at least once a week), how often they experienced cybervictimization (14 items, e.g., “had someone post insulting messages about you on a social networking site”) and engaged in cyberbullying (14 items, e.g., “posted insulting messages about someone on a social networking site”) in the past 12 months. Total scores were calculated by averaging items relevant to each subscale, with higher scores indicating more frequent experiences of cybervictimization and cyberbullying; Cronbach’s α = .93 and .95, respectively.

Social Desirability

The 13-item short Social Desirability Scale (Reynolds, 1982) was used to measure the tendency to engage in socially desirable responding. For each item (e.g., “I sometimes feel resentful when I don’t get my own way”), participants indicated true (1) or false (2). Items were averaged, with higher scores indicating more socially desirable responding; Cronbach’s α = .67.

Procedure

The study was advertised to students enrolled in a first-year psychology course. To compensate for participation, participants received course credit (0.5% of the total course grade) and were included in a prize draw. Participants accessed the online survey via a URL that directed them to a page describing their anonymity and their right to withdraw from the study at any time. Following a review of this page, participants were directed to the survey, which took about 20 min to complete.

Data Analytic Strategy

Small amounts of item level data were missing (range 0 to 3%) completely at random (Little’s MCAR test: χ2[2645] = 2734.66, p = .110). The expectation maximization algorithm in SPSS was used to replace the missing data in order to maintain all 270 participants in all analyses. Ms, SDs, and zero-order correlations between all measures and with gender were examined in preliminary analyses. For the primary analyses, we estimated all direct paths using AMOS v.26. In the hypothesized model, the focus was on estimating the paths from the measures of empathic distress, empathic anger, and compassion to the three bystander behavioral intentions in response to witnessing cyberbullying (the dependent variables of passive bystanding, aggressive defending, and prosocial defending). Gender, social desirability, history of bullying others, and history of personal victimization were included as covariates to account for their significant associations with independent and dependent variables. We report the findings in the final model after trimming nonsignificant paths.

Results

Correlations

Pearson’s correlations were conducted to examine the strength of the relationships among all measures and as preliminary tests of the hypotheses (see Table 2). Overall, the hypotheses were partially supported. With regard to passive bystanding, it was significantly and positively associated with empathic distress (p = .038) and significantly and negatively associated with empathic anger and compassion (p = .002 and .009, respectively). With regard to aggressive defending, associations with empathic anger (p = .208) and empathic distress (p = .390) were not significant. Yet, there was a significant negative correlation between aggressive defending and compassion (p = .024). Prosocial defending had significant positive associations with empathic anger and compassion (p < .001), but not with empathic distress (p = .263). Furthermore, bystander behavioral intentions were intercorrelated with each other, as were empathic distress, empathic anger, and compassion measures, and all covariates were associated with bystander behavioral intentions or empathic and/or compassionate constructs (see Table 2).

Table 2 Pearson correlations between all measures and means (M) and standard deviations (SD) of all measures (N = 270)
Full size table

Empathic Distress, Empathic Anger, and Compassion: Direct Associations with Bystander Behavioral Intentions in Response to Witnessing Cyberbullying

The first model tested direct pathways linking empathic distress, empathic anger, and compassion, as well as gender, social desirability, bullying, and victimization as covariates, with bystander behavioral intentions in response to witnessing cyberbullying. This model had an adequate fit to the data, \(\chi ^{2}\) (24) = 62.95, p < .001, CFI = 0.91; RMSEA = 0.078 (90% CI 0.055–0.101), p = .026. Because some of the covariates (i.e., social desirability and bullying) were not related to any other variables in this full model, we tested a reduced model, including only the covariates that were associated with at least one IV (i.e., empathic distress, empathic anger, compassion) and at least one DV (i.e., passive bystanding, aggressive defending, and prosocial defending).

Specifically, this reduced model tested the hypothesized direct pathways linking empathic distress, empathic anger, and compassion, as well as gender and victimization as covariates, with bystander behavioral intentions (i.e., passive bystanding, aggressive defending, and prosocial defending) in response to witnessing cyberbullying (see Table 3). This model had a good fit to the data, \(\chi ^{2}\) (9) = 14.61, p = .102, CFI = 0.98; RMSEA = 0.048 (90% CI 0.000–0.091), p = .474 (see Fig. 1), and accounted for 8% of the variance in passive bystanding, 11% of the variance in aggressive defending, and 25% of variance in prosocial defending. As hypothesized, significant direct paths were found for empathic distress to more passive bystanding and empathic anger to less passive bystanding, empathic anger to more aggressive defending and compassion to less aggressive defending, and both empathic anger and compassion to more prosocial defending. Contrary to H1c, the path from compassion to passive bystanding was not significant. Regarding covariates included in the model, gender was associated with aggressive defending, with male participants higher in aggressive defending compared to females. In addition, personal victimization was positively associated with aggressive defending.

Table 3 Unstandardized (B) and standardized (β) direct associations (effects) of empathic distress, empathic anger, and compassion, and covariates, with bystander behavioral intentions (passive bystanding, aggressive defending, prosocial defending) in response to witnessing cyberbullying (N = 270)
Full size table
Fig. 1
figure 1

Significant associations in the hypothesized model (covariates not depicted) (also see Table 3). Participant gender and history of personal cybervictimization were included as control variables; significant associations for covariates are shown in Table 3, and all nonsignificant paths for gender and cybervictimization were removed from the final model. Empathic distress and compassion were not significantly correlated, r = .07, p = .264, so this path was removed from the model. *p < .05. **p < .01. ***p < .001

Full size image

Discussion

In the present study, we hypothesized that there would be differential and unique associations of empathic distress, empathic anger, and compassion with intentions to engage in passive bystanding, aggressive defending, and prosocial defending in response to witnessing cyberbullying. Our broader purpose was to encourage the refinement of empathy-related intervention content and enhance bystander intervention effectiveness. While previous research has mainly focused on the importance of empathy for bystander defending (Lambe et al., 2019), we investigated whether it would be worthwhile to extend the focus on empathy in general to instead consider different empathy-related responses, including empathic distress, empathic anger, and compassion, that can emerge when witnessing bullying. After controlling for gender, cybervictimization, cyberbullying, and social desirability, most hypotheses were supported. Supporting our first hypothesis, students higher in empathic distress and lower in empathic anger reported that they would be more passive when they witnessed cyberbullying. However, in contrast to predictions, compassion was not uniquely associated with lower passive bystanding intentions. Supporting our second hypothesis, those higher in empathic anger and lower in compassion reported greater aggressive defending intentions. Finally, supporting our third hypothesis, those higher in empathic anger and compassion reported greater prosocial defending intentions.

Empathic Distress is Uniquely Associated with Passive Bystanding Intentions

The finding that students higher in empathic distress reported greater passive bystanding intentions is consistent with research and theory (e.g., Batson et al., 2015; Singer & Klimecki, 2014) suggesting that feelings of distress are associated with more self-protective behaviors (e.g., avoidance) when confronted with the distress of others. Hence, the focus of those high in empathic distress may be to reduce their own aversive feelings or avoid even more aversive feelings (e.g., being judged by others), rather than to defend. Indeed, empathic distress was not associated with either intentions to engage in aggressive or prosocial defending if witnessing cyberbullying. Moreover, consistent with the empathy-altruism hypothesis (Batson et al., 1983, 2015), empathic distress in response to witnessing the victimization of others may also indicate that a witness views the costs of intervening as too high, resulting in passivity rather than action and helping. However, future research is necessary to better understand whether the perceived costs of defending account for the association between empathic distress and passive bystanding.

The Differential and Unique Roles of Empathic Anger and Compassion

The finding that students higher in empathic anger reported lower passive bystanding intentions is consistent with previous bullying research focused on face-to-face interactions (Pozzoli et al., 2017) and supports Hoffman’s (2001) theory of empathy, suggesting that feelings of anger mobilize action that, in turn, hinders passivity. Relatedly, as expected, our results also revealed that empathic anger uniquely predicted intentions to use aggressive and prosocial defending in response to witnessing cyberbullying. These findings build on past research that has highlighted empathic anger as a moral motivator of both punishing behaviors directed at the person who offends and victim-directed helping behaviors (e.g., Vitaglione & Barnett, 2003).

In contrast to our hypothesis that compassion would uniquely predict lower passive bystanding intentions, the negative bivariate correlation between compassion and passivity was no longer significant when accounting for empathic distress and empathic anger in the path model. In part, this might reflect the complex nature of associations between the empathic and compassionate constructs. For example, Hoffman’s (2001) theory states that empathic concern is a prerequisite for the activation of empathic anger that in turn evokes action and inhibits passive bystanding. A recent cross-sectional study regarding bystander responses to in-person bullying supports this proposition, with self-reported feelings of empathic concern for other’s suffering (i.e., a component of compassion) having an indirect effect on passive bystanding via empathic anger (Pozzoli et al., 2017). To further investigate possible causal pathways such as this, future experimental and longitudinal research is needed.

As expected with regard to defending, our findings provide the first evidence of compassion as uniquely predicting lower aggressive defending intentions and greater prosocial defending intentions. This is consistent with research in the broader compassion literature suggesting that people who feel compassion are less likely to punish individuals who offend (e.g., McCall et al., 2014). Thus, the compassionate attribute of distress tolerance, which facilitates the ability to accept and tolerate difficult emotions (e.g., anger) that arise when witnessing the unfair treatment of a person being victimized, might help to prevent an aggressive bystander response (Gu et al., 2017; Stevens & Taber, 2021). In addition, because compassion is typically accompanied by greater feelings of concern for all people, even disliked others (Oveis et al., 2010; Sprecher & Fehr, 2005), compassion might facilitate a non-judgmental empathic view of the person who bullies that, in turn, prevents aggressive defending.

Further, the positive association between compassion and prosocial defending intentions aligns with the broader literature of prosocial behavior (e.g., Klimecki, 2019; Stevens & Taber, 2021), indicating that the compassionate attribute of distress tolerance of difficult empathic emotions also enables an attentional focus on the person being victimized. This is consistent with the empathy-altruism hypothesis (Batson et al., 1983, 2015) and the idea that other-oriented attentional focus, as opposed to self-oriented focus, is critical for constructive helping. Taken together then, while empathic anger is uniquely associated with a greater possibility of both forms of defending, compassion is uniquely associated with lower aggressive defending intentions and greater prosocial defending intentions.

Gender and Cybervictimization are Associated with Defending Behavioral Intentions

We included control variables in the multivariate path model to examine associations of empathic distress, empathic anger, and compassion with bystander behavioral intentions after partialling out the effects of gender and personal history of victimization established in past research (e.g., Bussey et al., 2020; Lambe & Craig, 2020). Of the controls included, participants’ gender, as well as their history of cybervictimization, were uniquely associated with greater aggressive defending intentions, but were not associated with prosocial defending or passive bystanding intentions. Intentions to engage in aggressive defending were more likely among males than females. Thus, even if the underlying motivation for defending is prosocial, males may perceive aggressive defending as a more socially acceptable strategy with fewer costs than females, consistent with how they are typically socialized to express more aggressive behavior than females (see Archer, 2019, for a review).

Practical Implications

In general, the findings of the current study support empathic distress, empathic anger, and compassion as differential correlates of bystander behavioral intentions in response to witnessing cyberbullying. Taken together these findings raise three considerations for interventions to potentially prevent less appropriate bystander behaviors (i.e., passive bystanding and aggressive defending) and promote more suitable harm reduction responses (i.e., prosocial defending). Firstly, given the positive association between empathic distress and passive bystanding intentions, it is essential to understand whether empathy training strategies inadvertently increase empathic distress, and subsequently, more passivity, rather than action, after witnessing bullying of others. Empathy training is designed to increase the ability to resonate with another’s distress, and, consequently, it may promote personal distress alongside recognition of the distress of the other and, ultimately, generate more passivity and avoidance (Singer & Klimecki, 2014). In contrast, bystanders who reported more other-oriented emotions in the current study, including empathic anger and compassion, were less likely to intend to remain passive. Hence, empathy interventions may benefit from including strategies to facilitate bystander’s ability to focus on the person who is victimized to potentially facilitate more approach-related helping behaviors (Singer & Klimecki, 2014).

A second aim of interventions should be to reduce bystanders’ use of aggressive defending strategies to avoid additional conflict and aggression and, rather, focus on ways to increase prosocial intervention that facilitates constructive conflict resolution. Although empathic anger predicts greater aggressive defending intentions, this moral empathic feeling might be critical to mobilize the action to intervene. Thus, rather than trying to diminish bystanders’ feelings of anger in interventions, interventions could focus on helping participants to learn skills and strategies that enable them to notice, regulate, and direct their anger to prosocial defending strategies. This might be particularly important for males who may be more likely to use aggressive defending. One way to channel anger might be to increase compassion, because compassion predicts lower intentions to aggressively defend and greater intentions to prosocially defend. Thus, compassion training should be considered when the aim is to facilitate constructive prosocial defending strategies. This is consistent with the broader compassion literature, with research showing that compassion training (e.g., loving-kindness meditation) is related to an increased concern for all human beings, altruistic helping, and a decreased desire to use punishment (Klimecki, 2019; McCall et al., 2014). Yet, to inform interventions, more research is needed to investigate causal relationships and the differential effects of empathy-based training versus compassion-based training across bullying contexts.

Limitations and Considerations for Future Research

The findings identified the complex associations of empathic and compassionate responses with bystander behavioral intentions in response to witnessing cyberbullying, but there are limitations in the study design. Firstly, we relied on self-reported dispositional measures of empathic distress, empathic anger, and compassion. Given that these constructs involve the experience of emotions, which are typically brief and context-specific, future research should investigate the extent to which situational/state-like empathic and compassionate reactions to different episodes of bullying translate to action in that same episode. Secondly, our measures were of bystander behavioral intentions in response to witnessing bullying, which may or may not reflect actual defending behaviors. Although there are theoretical grounds for believing that behavioral intentions predict behavior (e.g., the theory of planned behavior; Ajzen, 2011), and some research evidence is available to show a positive link between bystander behavioral intentions and actual defending behaviors (e.g., Banyard, 2008; DeSmet et al., 2016; McMahon et al., 2015), further research is needed to examine the extent to which empathic and compassionate reactions predict actual defending behaviors if students were to witness bullying in their day-to-day online interactions with others.

Thirdly, the cross-sectional design of our study does not allow for causal interpretation of the effects of empathic and compassionate responses on bystander behaviors. Moreover, while this study accounted for some variance in bystander behavioral intentions, there remains a large proportion of unaccounted for variance. Given the reduced contextual, social, and emotional cues available in an online bullying context, future research is needed to better understand whether the activation of empathic distress, empathic anger, and compassion is greater when witnessing in-person bullying. Finally, the current sample included young adult Australian university students with a higher proportion of females than males. The results may not generalize to other groups. Although our sample had a diverse ethic/racial background, the majority was white Australian. Given that minority groups (e.g., sexual, ethnic, individuals with disabilities) are at higher risk of victimization (e.g., Xu et al., 2020), it is critical to also investigate the associations of empathic and compassionate responses with different bystander behaviors across different ethnic-cultural groups and minorities.

Conclusion

Building upon previous research, this study demonstrates that empathic and compassionate constructs uniquely and differentially relate to intentions to engage in passive bystanding, aggressive defending, and prosocial defending in response to witnessing cyberbullying. Taken together, the results indicate that passivity in response to witnessing bullying is more likely intended if bystanders experience high levels of empathic distress and low levels of empathic anger. In terms of defending behaviors, empathic anger was also uniquely associated with greater aggressive and prosocial defending intentions, meaning that it is important to understand ways to direct anger to prosocial, constructive defending strategies. Finally, compassion was uniquely associated with lower aggressive defending intentions and greater prosocial defending intentions, making it unique in differentiating these two forms of defending. Future research is needed to investigate whether compassion training can enhance constructive bystander behaviors. Overall, our findings emphasize the presence of complex relationships between empathic and compassionate factors and bystander behavioral intentions in response to witnessing cyberbullying. The conclusions drawn from this study will be strengthened when the findings are replicated longitudinally and experimentally using situational measurements across bullying contexts and for more diverse populations.