Sie sind auf Seite 1von 10

Computers in Human Behavior 96 (2019) 186–195

Contents lists available at ScienceDirect

Computers in Human Behavior


journal homepage: www.elsevier.com/locate/comphumbeh

Full length article

Through the magnifying glass: Empathy's differential role in preventing and T


promoting traditional and cyberbullying
Daniel Graf∗, Takuya Yanagida, Christiane Spiel
University of Vienna, Faculty of Psychology, Department of Applied Psychology: Work, Education, and Economy, Universitätsstraße 7, 1010, Vienna, Austria

ARTICLE INFO ABSTRACT

Keywords: Empathy is considered a common protective factor against traditional and cyberbullying. Existing studies have
Bullying only considered broad components of empathy, such as affective and cognitive empathy. However, there are still
Cyberbullying inconclusive results regarding the role of cognitive empathy in (cyber)bullying. Therefore, we examined re-
Cognitive empathy lationships between not only components but also subcomponents of empathy and traditional and cyberbullying
Affective empathy
in order to investigate contextual differences (face-to-face, cyberspace) and unravel the blurry picture regarding
cognitive empathy. A total of 521 students (37.4% girls; Mage = 17.83 years; SD = 2.13) answered ques-
tionnaires during regular school hours on their traditional and cyberbullying involvement, empathy and cov-
ariates (age, gender, social media use, migration background and gaming attitudes). The results for cognitive
empathy revealed no differences in its protective function across contexts. However, the strength of the pro-
tective association was stronger for traditional than for cyberbullying. In contrast, affective empathy was not
related to either form of bullying. Subcomponents of cognitive empathy exhibited positive and negative asso-
ciations with (cyber)bullying. Considering subcomponents of affective empathy revealed differential relation-
ships with traditional and cyberbullying. The functional and quantitative differences in empathy's relation to
bullying across contexts and the results' potential contribution to the development of environment-specific in-
terventions are discussed.

1. Introduction preventing (cyber)bullying is high empathy (Antoniadou & Kokkinos,


2018; Jolliffe & Farrington, 2006). However, its association with
Bullying is defined as intended repetitive aggressive behavior that is (cyber)bullying is still unclear (van Noorden, Haselager, Cillessen, &
characterized by an imbalance of power (Olweus, 1993; Smith & Bukowski, 2015). The inconclusive findings may result from dis-
Ananiadou, 2003). In recent years, with the spread of new information regarding empathy's multifaceted nature on the one hand and from
and communication technologies (ICTs), new practices of engaging in neglecting differences between different forms of bullying (e.g., face-to-
such aggressive behavior have emerged (e.g., editing and publishing face, cyberbullying) on the other hand (van Noorden et al., 2015).
embarrassing pictures and videos). When considering these new prac- Empathy has been suggested to play both a common role (e.g., Casas,
tices, research largely follows the knowledge gained from traditional Del Rey, & Ortega-Ruiz, 2013; Del Rey et al., 2016) and differential
bullying research (Olweus, 2013) and defines bullying through ICTs as roles (e.g., Runions, 2013; Runions & Bak, 2015, Sutton, Smith, &
cyberbullying (Smith, Mahdavi, Carvalho, Fisher Rusell, & Tippett, Swettenham, 1999, 2001) across contexts (face-to-face, cyberspace).
2008; Tokunaga, 2010; Vivolo-Kantor, Martell, Holland, & Westby, However, empirical investigations are lacking. Therefore, the aim of the
2014). Nevertheless, there is no consensus as to whether bullying in the present study was to investigate the relationships between not only
two contexts (face-to-face, cyberspace) can be considered in a similar components but also subcomponents of empathy and traditional and
way (Olweus, 2012). cyberbullying. Furthermore, we investigated whether empathy plays a
Given that both forms of bullying entail negative consequences for differential role for bullying across contexts (face-to-face, cyberspace).
mental and physical health (e.g., Juvonen & Gross, 2008), examining To the best of our knowledge, this is the first study to examine sub-
protecting factors is crucial to provide information for prevention and components of affective and cognitive empathy in conjunction with
intervention programs. One widely considered protective factor for cyber- and traditional bullying.


Corresponding author. University of Vienna, Faculty of Psychology, Department of Applied Psychology: Work, Education and Economy, Universitätsstraße 7,
1010, Vienna, Austria.
E-mail address: daniel.graf@univie.ac.at (D. Graf).

https://doi.org/10.1016/j.chb.2019.02.007
Received 21 September 2018; Received in revised form 30 January 2019; Accepted 7 February 2019
Available online 11 February 2019
0747-5632/ © 2019 Elsevier Ltd. All rights reserved.
D. Graf, et al. Computers in Human Behavior 96 (2019) 186–195

2. Theoretical background conceptualization highlights the multifaceted nature of empathy, fol-


lowing the call by Decety (2011) to define empathy more precisely.
Research on aggression usually examines personal (e.g., traits, sex, They consider cognitive empathy to be a working model that can be
beliefs and attitudes) and situational (e.g., aggressive cues, provocation, continuously updated on the basis of information such as visual, audi-
frustration) factors as antecedents of aggressive behavior (Anderson & tory and situational cues. Furthermore, they distinguish between an
Bushman, 2002). The insights gained are then used to develop inter- intuitive component of cognitive empathy (perspective taking) and the
ventions with the aim of reducing aggressive behavior. The role of high more effortful projecting of others' emotions (online simulation; Reniers
empathy as a personal factor preventing aggressive behavior is broadly et al., 2011). Affective empathy, on the other hand, is defined as a
accepted (Antoniadou & Kokkinos, 2018; Jolliffe & Farrington, 2006): process elicited immediately by the rapid perception of emotional cues
Understanding and vicariously feeling the emotions of others circum- such as facial expressions, body gestures, and voice prosody, and that
vents behavior that is thought to harm another person (Eisenberg & evokes an emotional response reflecting another person's emotional
Miller, 1987). As a result, empathy trainings are implemented in many state (Reniers et al., 2011). This approach relies more on a person's
prevention and intervention programs tackling aggression in different ability to be sensitive to and vicariously experience the emotions of
settings, such as schools, correctional settings and families (Vachon, others, rather than being aware of them. According to Reniers et al.
Lynam, & Johnson, 2014). Nevertheless, despite their high cost, such (2011), affective empathy consists of three subcomponents: automatic
trainings do not always work as desired (Vachon et al., 2014; van mirroring of others' emotional states (emotional contagion), the sensi-
Noorden et al., 2015). Hence, there is a need to revisit the role of tivity to experience affective responses when witnessing others' emo-
empathy in aggressive behavior such as (cyber)bullying in order to tions in a close social context (proximal responsivity) and the sensitivity
enhance the effectiveness of empathy trainings implemented in pre- to vicariously experience the feelings of another person in a remote
vention and intervention programs. social context, for example when watching a film (peripheral re-
sponsivity; Reniers et al., 2011). At this point, it should be mentioned
2.1. The role of empathy in bullying research that the construct of empathy can be further characterized by the ability
to differentiate oneself from another person (self-other differentiation;
Several studies have examined the relationships between empathy Decety & Meyer, 2008). This allows for the self-aware separation of
and cyber- and/or traditional bullying (e.g., Ang & Goh, 2010; Del Rey oneself from others and includes the ability to acknowledge that the
et al., 2016; Jolliffe & Farrington, 2006; Steffgen, König, Pfetsch, & source of the shared emotion lies with another person (Decety &
Melzer, 2011). These studies apply a common conceptualization of Jackson, 2004). In this sense, emotion contagion in and of itself should
empathy as a personality trait consisting of affective and cognitive be regarded as a precursor of affective empathy. This is because self-
components (e.g., Decety & Jackson, 2006; Jolliffe & Farrington, 2006; other differentiation is not considered when defining emotion con-
Walter, 2012; Zaki & Ochsner, 2012). This conceptualization describes tagion. However, the lower one's ability to differentiate oneself from
affective empathy as bottom-up processing involving affect sharing or another person, the more overwhelming one's own emotions will be
vicariously experiencing the emotions of others. On the other hand, when witnessing someone else in a stressful situation (Decety & Meyer,
cognitive empathy is seen as top-down processing that involves un- 2008). This, in turn, may lead one's focus to shift away from the other
derstanding others' emotions, including the motivational, intentional person and towards oneself, which is no longer consistent with the
and self-regulative tailoring of empathic experiences. Empathic ex- definition of empathy (Decety & Meyer, 2008).
periencing is considered a result of both bottom-up and top-down Nevertheless, a differentiated view on empathy's multifaceted
processing (for more details see Decety & Lamm, 2006; Decety, 2011). nature allows us to gain insights into the blurry picture regarding the
A large body of existing studies have found negative relationships relationship between cognitive empathy and (cyber)bullying. By con-
between both components of empathy and traditional and cyberbul- sidering two contexts (face-to-face, cyberspace), we are further able to
lying. For example, a recent meta-analysis found that cyberbullies had investigate differences in the strength and/or direction of the re-
higher odds of scoring low in both cognitive and affective empathy. lationships between components and subcomponents of empathy and
However, the authors stated that there were not enough studies to draw traditional and cyberbullying.
firm conclusions (Zych, Baldry, Farrington, & Llorent, 2018). Further-
more, in a systematic review, van Noorden et al. (2015) reported that 2.3. Contextual differences between cyber- and traditional bullying
bullying in general, without specifying the context, is negatively asso-
ciated with affective and cognitive empathy. Moreover, the relationship There are a number of substantial differences in the contextual
between bullying and affective empathy seemed clearer than the re- properties of cyber- and traditional bullying that must be considered
lationship between bullying and cognitive empathy (van Noorden et al., (e.g., Runions, 2013; Runions & Bak, 2015; Suler, 2004). Computer-
2015). Nevertheless, the authors did not distinguish between cyber- and mediated communication (CMC) offers the possibility to communicate
traditional bullying, discussing the effects across contexts. anonymously. This means that a significant share of victims do not
Results concerning the relationship between cognitive empathy and know or recognize their perpetrators in cyberspace (Kowalski & Limber,
bullying are controversial (Vachon et al., 2014). For example, Sutton 2007). In addition to more unpleasant feelings for the victims in such
et al. (1999) criticize the common social skills deficit model, arguing anonymous situations (Mishna, Saini, & Solomon, 2009; Sticca &
that understanding the emotions of others is useful for manipulating Perren, 2013), anonymity may facilitate aggressive behavior such as
and organizing other perpetrators and victims. He and other colleagues bullying due to the online disinhibition effect (Suler, 2004). More
suggest that the ability to adopt another person's perspective facilitates specifically, anonymous online communication may lead to a separa-
the achievement of the bully's goals by promoting the application of the tion of face-to-face behavior and online actions in which online per-
most suitable aggressive behavior in a specific social situation (e.g., petrators see their communication partner in cyberspace as more an
Sutton et al., 1999, 2001). In sum, more research is needed to unravel avatar rather than an actual human being (Suler, 2004; Terry & Cain,
contradictions regarding the relationship between cognitive empathy 2016). This is in line with social presence theory, which postulates that
and (cyber)bullying as well as a more differentiated construct of em- CMC affects the perceived salience of others (for a detailed explanation
pathy, as already called for by Decety (2011). see Short, Williams, & Christie, 1976; Gunawardena, 2004). In contrast,
the perceived social context should be closer in face-to-face commu-
2.2. A differentiated construct of empathy nication compared to CMC.
Regarding the influence of contextual properties on the relationship
Reniers, Corcoran, Drake, Shyrane, and Völlm's (2011) between empathy and (cyber)bullying, the social information

187
D. Graf, et al. Computers in Human Behavior 96 (2019) 186–195

processing theory of computer-mediated communication (SIP) is useful affective empathy and cyberbullying than between affective empathy
to rely on (Walther, 1992). According to the SIP, a lack of nonverbal and traditional bullying (Hypothesis 3). As we assumed that more
cues in computer-mediated communication (CMC) leads to a resources are needed in cyberspace to elicit cognitive empathy, we
prolonging of the function of communication compared to in a multi- hypothesized a difference in the strength of the relationships between
modal communication environment (Walther, 2011). In other words: cognitive empathy and bullying across the two contexts (Hypothesis 4).
more resources (e.g., time, cognitive capacities) are needed to generate Again, we did not specify a direction due to the existence of conflicting
a sufficient cognitive model of one's communication partners in CMC perspectives regarding cognitive empathy's function in (cyber)bullying.
(Walther, 2011). According to this assumption, understanding the Finally, we expected that the pattern of relations would be invariant
emotions of others (cognitive empathy) may also need more resources when statistically controlling for covariates (Hypothesis 5).
in a cyber, compared to a face-to-face context and therefore may be
harder to elicit in a cyber context. Furthermore, the lack of nonverbal 2.4.1.2. Subcomponents of affective and cognitive empathy. Guided by
cues (e.g., visual, auditory or situational) in the cyber context may hide our hypotheses concerning higher order components, we also
underlying emotional cues (e.g., facial expressions, body gestures, and hypothesized relationships between all subcomponents of affective
voice prosody) that may trigger affective empathy (Runions, 2013). empathy and bullying in both contexts (Hypothesis 6). Regarding the
Consequently, eliciting affective and cognitive empathic responses subcomponents of cognitive empathy, we assumed relationships
might be more difficult in the cyber context, with its paucity of non- between all subcomponents of cognitive empathy and both forms of
verbal, paralinguistic and semantic social cues (Runions, 2013). bullying without specifying a direction (Hypothesis 7). Furthermore,
we postulated differences in the strength of the relationships between
2.4. The present study subcomponents of cognitive empathy and bullying in the cyber and
face-to-face context (Hypothesis 8). We postulated these differences due
The aim of the present study was first to investigate differential to assumed differences in available resources (e.g., information to
relationships between affective and cognitive empathy and their re- generate a sufficient cognitive model of others) between contexts (face-
spective subcomponents with (cyber)bullying and second to examine to-face, cyberspace). Regarding affective empathy, we assumed a
differences in empathy's role in bullying across contexts (face-to-face, stronger negative relationship between peripheral responsivity and
cyberspace). To the best of our knowledge, this is the first study to cyberbullying than between peripheral responsivity and traditional
examine subcomponents of affective and cognitive empathy in con- bullying (Hypothesis 9) based on the online disinhibition effect (Suler,
junction with cyber- and traditional bullying. 2004) and social presence theory (e.g., Gunawardena, 2004; Short
By examining the relationships between subcomponents of cogni- et al., 1976). On the other hand, we hypothesized a stronger negative
tive empathy and bullying in two contexts (face-to-face, cyberspace), relationship between proximal responsivity and bullying in the face-to-
we expect to contribute to explaining contradicting results regarding face context than between proximal responsivity and cyberbullying
the relationship between cognitive empathy and (cyber)bullying. (Hypothesis 10). We assumed this due to theoretical considerations that
Moreover, by investigating to what extent components and sub- the perceived social context in face-to-face interaction is closer than in
components of empathy are similarly or differently related to cyber- CMC. Finally, we assumed that the results would be invariant when
and traditional bullying, the current study additionally contributes to statistically controlling for covariates (Hypothesis 11). We did not
the discussion of to what extent cyber- and traditional bullying should specify any hypotheses regarding contextual (face-to-face, cyberspace)
be considered in a similar way. We first examined relationships be- differences for emotion contagion.
tween components of empathy and both forms of bullying by using the
frequently applied conceptual distinction between cognitive and af- 3. Methods
fective empathy (Model 1), without considering their subcomponents.
We then added age, gender, social media use, migration background 3.1. Sample and procedure
and gaming attitudes as covariates to test the robustness of the observed
pattern of relationships in Model 1 (Model 2). Subsequently, we sub- Students from four Austrian vocational schools participated in the
divided cognitive and affective empathy into their subcomponents as study. They answered online questionnaires in their school's own
suggested by Reniers et al. (2011) for a more detailed analysis (Model computer lab during regular school hours. Trained research assistants
3). Finally, we again added the covariates to test the invariance of the were present at all times. Participation was voluntary, and participants
relation pattern observed in Model 3 (Model 4). In all models, we fur- provided informed consent. The consent rate was above 99%. Overall,
ther examined differences in the strength of the relationships between 521 students (37.4% girls; Mage = 17,83 years; SD = 2.13) partici-
components and subcomponents of empathy and cyber- and traditional pated. 89.1% of students reported that their most frequently used lan-
bullying. guage was German.

2.4.1. Hypotheses 3.2. Measures


In this section, we present our hypotheses, sorted by whether they
concern relations between components or subcomponents of empathy To test our previously selected questionnaire battery, we initially
and cyber- and traditional bullying. presented our survey to students falling within our target group. After
providing an introduction about (cyber)bullying, we asked for com-
2.4.1.1. Components of empathy: affective and cognitive empathy. Based ments regarding the comprehensibility of the questionnaire. Overall,
on previous studies (van Noorden et al., 2015; Zych et al., 2018), we the items were easy to understand. Only a few items were slightly
expected negative relationships between affective empathy and modified to better reflect the every-day language of adolescents without
bullying in the cyber as well as in the face-to-face context modifying the semantic content. These students were not part of the
(Hypothesis 1). As existing results are inconclusive regarding the actual sample used in this study.
relationship between cognitive empathy and (cyber)bullying, we
formulated an undirected hypothesis and stated that there would be a 3.2.1. Empathy
relationship between cognitive empathy and bullying in both contexts We measured the (sub)components of empathy using the German
(face-to-face, cyberspace) without specifying a direction (Hypothesis 2). version of the 31-item Questionnaire of Cognitive and Affective
Furthermore, as the lack of nonverbal cues in cyberspace hides Empathy (QCAE; Reniers et al., 2011), translated by Georgi, Petermann,
emotional cues, we expected a weaker negative relationship between and Schipper (2014). The QCAE is based on existing questionnaires

188
D. Graf, et al. Computers in Human Behavior 96 (2019) 186–195

measuring cognitive and/or affective empathy (e.g., The Empathy first covariate was gender, as research has repeatedly shown greater
Quotient; Baron-Cohen, Richler, Bisarya, Gurunathan, & Wheelwright, engagement in traditional bullying behavior among boys (Kowalski,
2003; The Hogan Empathy Scale; Hogan, 1969; The Interpersonal Re- Giumetti, Schroeder, & Lattanner, 2014) and higher empathy scores
activity Index; Davis, 1983). The QCAE encompasses a second-order among females (Baron-Cohen & Wheelwright, 2004). Furthermore, we
structure with two subcomponents related to cognitive empathy: a) included age as a covariate, as research has shown variations in the
perspective taking (10 items; example item: I can tell if someone is prevalence rates of traditional bullying and cyberbullying depending on
masking their true emotion) and b) online simulation (9 items; example age (Kowalski et al., 2014). To account for findings that cyberbullying
item: Before criticizing somebody, I try to imagine how I would feel if I may be a harmful effect of social media use (Best, Manktelow, & Taylor,
was in their place). Three subcomponents were related to affective 2014), we asked participants how often they check social media right
empathy: a) emotion contagion (4 items; example item: I am happy after getting up on a 5-point response scale (1 = never, 2 = rarely,
when I am with a cheerful group and sad when the others are glum), b) 3 = sometimes, 4 = often, 5 = always). As studies suggest that video
proximal responsivity (4 items; example item: It affects me very much games affect empathy and aggressive behavior (e.g., Anderson et al.,
when one of my friends seems upset), and c) peripheral responsivity (4 2010; Greitemeyer & Mügge, 2014), we asked participants to what
items; example item: I often get deeply involved with the feelings of a extent they would describe themselves as a gamer on a 5-point response
character in a film, play, or novel). On a 4-point Likert scale (1 = no, scale (1 = not at all, 2 = a little, 3 = average, 4 = fairly, 5 = very
2 = somewhat no, 3 = somewhat yes, 4 = yes), participants had to much). Finally, we measured migration background by asking partici-
agree or disagree with each statement. The QCAE has been found to pants if their main language at home is German (i.e. the official national
exhibit high internal consistency and good validity (Georgi et al., 2014; language of Austria), since several studies have suggested higher levels
Reniers et al., 2011). Assuming that a shorter questionnaire has a of traditional bullying among immigrants (e.g., Fandrem, Strohmeier, &
higher probability of participants answering the items reliably and Roland, 2009; Strohmeier, Fandrem, & Spiel, 2012).
without dropout, we shortened the questionnaire battery by using four
items per subcomponent. To do so, we selected the four items with the 3.3. Missing data
strongest factor loadings on perspective taking and online simulation
according to Reniers et al. (2011), resulting in 20 items overall. As In total, 0.08% of data were missing, stemming from 20 incomplete
shown in Table 1, internal consistency was good for cognitive (ω records. The mean percentage of missing values across the 53 variables
= .81) and affective empathy (ω = .84). However, perspective taking ranged between 0.00% and 2.87%. Pairwise deletion based on the de-
(ω = .62), emotion contagion (ω = .61), and peripheral responsivity fault setting for using the robust weighted least squares estimator
(ω = .69) had an internal consistency below .70. (WLSMV) in Mplus (Muthén & Muthén, 1998–2018) was applied. Note
that pairwise deletion is not recommended for handling missing data in
general (Enders, 2010), but defensible given the small amount of
3.2.2. Cyberbullying and traditional bullying
missing data.
We assessed self-reported cyberbullying and traditional bullying
behavior using The European Cyberbullying Intervention Project
3.4. Measurement models
Questionnaire (ECIPQ; Del Rey et al., 2015). Participants had to in-
dicate whether they had intentionally engaged in cyberbullying or
First, measurement models based on a first-order factor model for
traditional bullying behavior within the last 2 months. On a 5-point
the subcomponents of cognitive and affective empathy were estab-
response scale (1 = no, 2 = yes, one or two times, 3 = yes, one or two
lished. Second, measurement models based on a second-order factor
times per month, 4 = yes, approximately one time per week, 5 = yes,
model for cognitive and affective empathy in accordance with the hy-
more than once a week) students reported how often they participated
pothesis that the second-order factors cognitive and affective empathy
in such behavior (example item for cyberbullying: I hacked into
account for the relations among the five subcomponents of empathy
someone's account and stole personal information; example item for
were established. Last, measurement models for cyberbullying and
traditional bullying: I hit, kicked, or pushed someone). The cyberbul-
traditional bullying based on a one-factor model were established.
lying scale includes 11 items and the traditional bullying scale 7 items.
Confirmatory factor analysis (CFA, see Brown, 2016) was conducted
The ECIPQ has been structurally validated in six countries (Del Rey
in Mplus 8.1 (Muthén & Muthén, 1998–2018) to test the measurement
et al., 2015). As can be seen in Table 1, internal consistency was very
models. In order to take into account the ordered-categorical nature of
good for both cyberbullying (ω = .96) and for traditional bullying (ω
the scale items, CFA with ordered-categorical indicators was applied
= .93).
using the robust weighted least squares estimator (WLSMV, see Bovaird
& Koziol, 2012). Measurement models were evaluated using the fit in-
3.2.3. Covariates dices CFI, RMSEA, and SRMR and common cut-off criteria (see Kline,
Several variables for which relations to empathy and (cyber)bul- 2015).
lying have previously been observed were considered as covariates. The The results revealed a very good model fit of the first-order factor
model for cognitive empathy (CFI = .980, RMSEA = 0.040, and
Table 1 SRMR = 0.031), with standardized factor loadings ranging from .34 to
Empathy, cyberbullying and traditional bullying: Internal consistency. .71. The first-order factor model for affective empathy also exhibited a
Scale # items Ordinal Cronbach's α Ordinal McDonald‘s ω very good model fit (CFI = .969, RMSEA = 0.046, and SRMR = 0.043).
The standardized factor loadings, however, revealed a negative factor
Cognitive empathy 8 .72 .81 loading for an item measuring peripheral responsivity. Inspecting the
Perspective taking 4 .58 .62
item content (“It's hard for me to understand why some things make
Online simulation 4 .77 .78
Affective empathy 12 .80 .84 other people so upset”) revealed that its content contradicted the the-
Emotion contagion 4 .58 .61 oretical considerations introduced earlier. First, it contains the word
Proximal responsivity 4 .67 .75 “understanding”. Since understanding others' emotions is a crucial part
Peripheral responsivity 4 .48 .69 of cognitive empathy, this may have led to confusion. Second, it refers
Cyberbullying 11 .95 .96
Traditional Bullying 7 .89 .93
to “other people”. This does not match the content of the other items,
which measure peripheral responsivity by specifically referring to
Note. Ordinal Cronbach's α and McDonald‘s ω were computed in R (R Core persons in a film, novel or play. As a result, this item was omitted from
Team, 2018) using the psych package version 1.8.4 (Revelle, 2018). the measurement model for peripheral responsivity. Re-estimation of

189
D. Graf, et al. Computers in Human Behavior 96 (2019) 186–195

Table 2 by cognitive and affective empathy (Hypotheses 1 & 2). In Model 2, the
Confirmatory factor analysis results: Empathy, cyberbullying and traditional covariates gender, age, migration background, social media use, and
bullying. gaming attitudes were added and statistically controlled for
Scale χ² df CFI RMSEA SRMR (Hypothesis 5). In Model 3, cyberbullying and traditional bullying were
predicted by all five subcomponents of empathy (Hypotheses 6 and 7).
Cognitive empathy, first-order model 35.14 19 .980 0.040 0.031 In Model 4, all covariates were added and statistically controlled for
Perspective taking
(Hypothesis 11).
Online simulation
Affective empathy, first-order model 89.18 41 .975 0.047 0.040 In all models, differences between the face-to-face and cyber context
Emotion contagion in the regression slopes for cognitive and affective empathy as well as
Proximal responsivity the five facets of empathy were tested for statistical significance
Peripheral responsivity
(Hypotheses 3–4 and 8–10). To assure a common metric across cyber-
Cognitive empathy, second-order model 35.14 19 .980 0.040 0.031
Affective empathy, second-order model 89.18 41 .974 0.048 0.040
and traditional bullying, the effect coding method (Little, Slegers, &
Cyberbullying 77.53 41 .985 0.041 0.049 Card, 2006) was used to identify and scale the latent variables.
Traditional Bullying 31.75 11 .985 0.060 0.034 Statistical analyses were conducted using Mplus version 8.1
(Muthén & Muthén, 1998–2018). Models were estimated using the ro-
bust weighted least squares estimator (WLSMV). We accounted for the
the model showed an increase in model fit (CFI = .975, hierarchical data structure (i.e., students nested in classes) by adjusting
RMSEA = 0.047, and SRMR = 0.040), with standardized factor load- the standard errors using a sandwich estimator taking into account the
ings ranging from .32 to .79. The second-order factor model for cog- non-independence of observations.
nitive empathy exhibited a very good model fit (CFI = .980, All analyses were conducted based on a statistical significance level
RMSEA = 0.040, and SRMR = 0.031), with standardized factor load- α = .05.
ings of .64 for perspective taking and .74 for online simulation.
Likewise, the model fit for the second-order factor model for affective
4. Results
empathy exhibited a very good model fit (CFI = .974, RMSEA = 0.048,
and SRMR = 0.040), with standardized factor loadings of .99 for
4.1. Descriptive statistics
emotion contagion, .97 for proximal responsivity, and .66 for periph-
eral responsivity. The measurement model for cyberbullying exhibited
Correlation coefficients, means and standard deviations of all vari-
a very good model fit as well (CFI = .985, RMSEA = 0.041, and
ables are shown in Table 3. Cyberbullying was negatively related with
SRMR = 0.049), with standardized factor loadings ranging from .63 to
cognitive empathy (r = −.18), online simulation (r = −.23), affective
.93. Similarly, the measurement model for traditional bullying ex-
empathy (r = −.16), emotion contagion (r = −.14), and peripheral
hibited a very good model fit (CFI = .985, RMSEA = 0.060, and
responsivity (r = −.18). No statistically significant relations were
SRMR = 0.034), with standardized factor loadings ranging from .69 to
found for perspective taking and proximal responsivity. Traditional
.80. In sum, the results revealed very good model fit for all scales, in-
bullying was negatively correlated with cognitive empathy (r = −.26),
dicating that all scales had sound measurement properties (see Table 2).
online simulation (r = −.33), affective empathy (r = −.18), emotion
contagion (r = −.17), proximal responsivity (r = −.15) and peripheral
3.5. Analytic strategy responsivity (r = −.10). No statistically significant relation was found
for perspective taking.
A structural equation modeling (SEM) approach based on the
measurement models with ordered-categorical indicators was used to 4.2. Relations between affective and cognitive empathy and cyberbullying
test the study's main hypotheses (see Kline, 2015). More specifically, and traditional bullying (Model 1 and Model 2)
cyber- and traditional bullying were predicted using components and
subcomponents of empathy in order to investigate the proposed re- Model 1 investigating the relationship between affective and cog-
lationships. nitive empathy and cyberbullying and traditional bullying (see Table 4)
In Model 1, cyberbullying and traditional bullying were predicted exhibited a good model fit (χ²(608) = 832.30, CFI = .941,

Table 3
Descriptive statistics: Bivariate Correlations, Means and Standard Deviations.
Variable 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

1. Cognitive empathy
2. Perspective taking .73
3. Online simulation .87 .29
4. Affective empathy .43 .28 .41
5. Emotion contagion .29 .16 .28 .83
6. Proximal responsivity .46 .31 .41 .81 .55
7. Peripheral responsivity .30 .20 .27 .75 .41 .38
8. Cyberbullying -.18 -.04 -.23 -.16 -.14 -.06 -.18
9. Traditional bullying -.26 -.04 -.33 -.18 -.17 -.15 -.10 .62
10. Gender -.12 -.08 -.11 -.39 -.23 -.35 -.35 .04 .07
11. Age .02 -.02 .04 -.01 .01 -.08 .04 -.04 -.04 .19
12. Migration background .00 .01 -.01 .02 .00 .05 .02 .09 .11 .04 .08
13. Social media -.05 .07 -.12 .09 .10 .11 .01 .14 .10 -.16 -.08 .04
14. Gamer -.03 .00 -.04 -.17 -.14 -.15 -.11 .04 .01 .37 .04 -.02 -.05

M 2.99 3.20 2.78 2.84 2.74 3.03 2.71 1.19 1.36 0.63 17.83 0.11 3.25 2.70
SD 0.46 0.48 0.66 0.50 0.59 0.57 0.76 0.38 0.49 2.13 1.39 1.47

Note. N = 523; Gender is coded as 0 = females and 1 = males; Migration background is coded as 0 = German and 1 = Other language; Statistically significant results
at α = .05 are in boldface.

190
D. Graf, et al. Computers in Human Behavior 96 (2019) 186–195

Table 4
Structural equation modeling results: Cyberbullying and traditional bullying on empathy.
Model Cyberbullying Traditional Bullying

Est. (SE) Std. Est. Est. (SE) Std. Est.

Model 1: χ²(608) = 832.30, CFI = .941, RMSEA = 0.027, SRMR = 0.073


Cognitive empathy −1.13 (0.30) −0.36 −1.50 (0.32) −0.50
Affective empathy 0.01 (0.10) 0.01 0.10 (0.11) 0.07

Model 2: χ²(773) = 1034.76, CFI = .927, RMSEA = 0.025, SRMR = 0.074


Cognitive empathy −1.00 (0.30) −0.31 −1.47 (0.34) −0.49
Affective empathy −0.03 (0.15) −0.02 0.13 (0.15) 0.09
Gender 0.01 (0.14) 0.01 0.14 (0.11) 0.09
Age −0.02 (0.02) −0.04 −0.02 (0.02) −0.05
Migration background 0.21 (0.13) 0.08 0.27 (0.12) 0.11
Social media use 0.08 (0.02) 0.15 0.05 (0.02) 0.10
Gaming attitudes 0.05 (0.03) 0.09 −0.00 (0.03) −0.01

Model 3: χ²(598) = 833.25, CFI = .928, RMSEA = 0.027, SRMR = 0.073


Perspective taking 0.44 (0.14) 0.28 0.44 (0.09) 0.29
Online simulation −0.49 (0.12) −0.43 −0.60 (0.11) −0.56
Emotion contagion −0.01 (0.08) 0.00 0.10 (0.08) 0.00
Proximal responsivity 0.16 (0.10) 0.12 −0.15 (0.15) −0.12
Peripheral responsivity −0.28 (0.10) −0.25 0.01 (0.13) 0.01

Model 4: χ²(748) = 1040.86, CFI = .918, RMSEA = 0.027, SRMR = 0.071


Perspective taking 0.28 (0.14) 0.18 0.33 (0.10) 0.22
Online simulation −0.49 (0.11) −0.42 −0.61 (0.11) −0.56
Emotion contagion 0.07 (0.08) 0.00 0.15 (0.07) 0.01
Proximal responsivity 0.19 (0.11) 0.14 −0.11 (0.15) −0.08
Peripheral responsivity −0.33 (0.10) −0.29 −0.01 (0.12) −0.01
Gender −0.04 (0.15) −0.03 0.11 (0.10) 0.08
Age −0.00 (0.02) −0.01 −0.02 (0.01) −0.05
Migration background 0.19 (0.14) 0.07 0.26 (0.12) 0.11
Social media use 0.05 (0.03) 0.09 0.03 (0.03) 0.05
Gaming attitudes 0.05 (0.04) 0.09 −0.01 (0.03) −0.02

Note. Est. = Unstandardized estimate; SE = Standard error; Std. Est. = Standardized Estimate; Gender is coded as 0 = females and 1 = males; Migration background
is coded as 0 = German and 1 = Other language; Statistically significant results at α = .05 are in boldface.

RMSEA = 0.027, SRMR = 0.073). As expected, the results revealed a Model 2 investigating the relationship between cyberbullying and
negative relationship between cognitive empathy and cyberbullying ( ˆ traditional bullying and affective and cognitive empathy while con-
= −1.13, p < .001) as well as traditional bullying ( ˆ = −1.50, trolling for covariates (see Table 4) exhibited a good model fit (χ²(773)
p < .001) while statistically controlling for affective empathy. Affec- = 1034.76, CFI = .927, RMSEA = 0.025, SRMR = 0.074). The results
tive empathy was not related to either cyberbullying ( ˆ = 0.01, indicated a positive relationship between social media use and cyber-
p = .930) or traditional bullying ( ˆ = 0.10, p = .343) when statistically bullying ( ˆ = 0.08, p < .001) and a positive relationship between
controlling for cognitive empathy. The difference in regression slopes migration background ( ˆ = 0.27, p = .021) and social media use and
was statistically significant for cognitive empathy ( ˆ = −0.38, traditional bullying ( ˆ = 0.05, p = .037). All other covariates were
p = .032), but not for affective empathy ( ˆ = 0.01, p = .188). That is, statistically insignificant. Statistically controlling for all covariates and
the negative relationship between bullying and cognitive empathy is affective empathy, there was a negative relationship between cognitive
stronger in the traditional context compared to the cyber context (see empathy and cyberbullying ( ˆ = −1.00, p < .001) as well as tradi-
Fig. 1, panel a). tional bullying ( ˆ = −1.47, p < .001). Affective empathy was not

(a) (b)

Gender
Traditional Cognitive Traditional
Cognitive -1.47 0.14
- 1.50 bullying empathy bullying
empathy 0.01
0.05
Social
0.27 media use
-1.13 -1.00 0.08

Migration
background

0.10 0.13 -0.00


0.21
Gaming
0.05 attitudes
-0.02
Affective Cyber- Affective -0.03 Cyber- -0.02
0.01 empathy bullying
empathy bullying Age

Fig. 1. Unstandardized estimates: Cyberbullying and traditional bullying on cognitive and affective empathy without and with covariates. Statistically significant
results at α = .05 are in boldface.

191
D. Graf, et al. Computers in Human Behavior 96 (2019) 186–195

Fig. 2. Unstandardized estimates: Cyberbullying and traditional bullying on subcomponents of cognitive and affective empathy without and with covariates.
Statistically significant results at α = .05 are in boldface.

related to cyberbullying ( ˆ = −0.03, p = .821) or traditional bullying −0.49, p < .001) and peripheral responsivity ( ˆ = −0.33, p < .001)
( ˆ = 0.13, p = .398) when statistically controlling for all covariates and cyberbullying and a positive relationship between perspective
and cognitive empathy. The difference in regression slopes was statis- taking and cyberbullying ( ˆ = 0.28, p = .040) while statistically con-
tically significant for cognitive empathy ( ˆ = −0.47, p = .010), but trolling for all other subcomponents of empathy and all covariates.
not for affective empathy ( ˆ = 0.16, p = .168). That is, the negative Furthermore, a negative relationship between online simulation and
relationship between bullying and cognitive empathy is also stronger in traditional bullying ( ˆ = −0.61, p < .001) and positive relationships
the traditional context compared to the cyber context when controlling between perspective taking ( ˆ = 0.33, p < .001) and emotion con-
for covariates (see Fig. 1, panel b). tagion ( ˆ = 0.15, p = .039) and traditional bullying were found while
statistically controlling for all covariates and all other subcomponents
of empathy. The relationships between all other subcomponents and
4.3. Relations between subcomponents of empathy and cyberbullying and
traditional bullying were statistically insignificant. The differences in
traditional bullying (Model 3 and Model 4)
regression slopes were statistically significant for proximal responsivity
( ˆ = −0.29, p = .017) and peripheral responsivity ( ˆ = 0.32,
Model 3 investigating the relationship between the subcomponents
p < .001), as expected. All other differences in regression slopes were
of empathy and cyberbullying and traditional bullying (see Table 4)
statistically insignificant (see Fig. 2, panel b).
exhibited a good model fit (χ²(598) = 833.25, CFI = .928,
RMSEA = 0.027, SRMR = 0.073). The results revealed negative re-
lationships between online simulation ( ˆ = −0.49, p < .001) and 5. Discussion
peripheral responsivity ( ˆ = −0.28, p = .004) and cyberbullying and a
positive relationship between perspective taking and cyberbullying ( ˆ Empathy is widely considered to be a protective factor preventing
= 0.44, p < .001) while statistically controlling for all other sub- (cyber)bullying (Antoniadou & Kokkinos, 2018; Jolliffe & Farrington,
components of empathy. The relationships between all other sub- 2006). However, its association with (cyber)bullying is still unclear
components and cyberbullying were statistically insignificant. Fur- (van Noorden et al., 2015). As the inconclusive findings may have re-
thermore, a negative relationship between online simulation and sulted from solely focusing on cognitive and affective empathy, dis-
traditional bullying ( ˆ = −0.60, p < .001) and a positive relationship regarding empathy's multifaceted nature, and from neglecting to take a
between perspective taking and traditional bullying ( ˆ = 0.44, different view of different forms of bullying (e.g., face-to-face, cyber-
p < .001) were found while statistically controlling for all other sub- bullying; e.g., van Noorden et al., 2015), the present study aims to
components of empathy. The relationships between all other sub- overcome these deficits. We investigated differential relationships be-
components and traditional bullying were statistically not significant. tween cognitive and affective empathy as well as their subcomponents
The differences in regression slopes were statistically significant for (in accordance with Reniers et al., 2011) and traditional and cyber-
proximal responsivity ( ˆ = −0.30, p = .014) and peripheral re- bullying while considering several covariates. Thus, the present study
sponsivity ( ˆ = 0.29, p = .004), as expected. All other differences in can provide deeper insights into contradicting results regarding the
regression slopes were statistically insignificant (see Fig. 2, panel a). relationship between cognitive empathy and (cyber)bullying and into
Model 4 investigating the relationship between cyberbullying and the associations between subcomponents of cognitive and affective
traditional bullying and the subcomponents of empathy while con- empathy and both forms of bullying. Below, we discuss the results
trolling for the covariates (see Table 4) exhibited a good model fit concerning the two components of empathy (affective and cognitive)
(χ²(748) = 1040.86, CFI = .918, RMSEA = 0.027, SRMR = 0.071). and their relations to cyber- and traditional bullying. In addition, we
For cyberbullying, the results showed a positive relationship with the discuss associations between subcomponents of affective and cognitive
covariate social media use ( ˆ = 0.05, p = .050), while all other cov- empathy and cyber- and traditional bullying. Next, we consider the
ariates were not significant. For traditional bullying, a positive re- relationships between covariates and bullying in both contexts (face-to-
lationship with migration background ( ˆ = 0.26, p = .029) was ob- face, cyberspace) and discuss the invariance of the results when adding
served. All other covariates were statistically insignificant. covariates. Finally, limitations, implications, a summary and conclu-
There was a negative relationship between online simulation ( ˆ = sions are reported.

192
D. Graf, et al. Computers in Human Behavior 96 (2019) 186–195

5.1. Components of empathy: affective and cognitive empathy support the conclusions drawn by Sutton et al. (1999, 2001) that the
ability to know the emotions of others can be exploited to harm others.
We found no significant relationships between affective empathy Our findings further suggest that the willingness and effort to adopt
and cyber- or traditional bullying. Furthermore, no differences were another person's perspective is a sufficient condition and a setscrew that
observed in the strength of these non-significant relationships between should be considered in prevention and intervention programs. Con-
contexts. This contradicts our hypotheses (Hypothesis 1 and Hypothesis trary to our assumptions, we found no differences in the strength of
3). Despite a significant body of research demonstrating the relation- these relationships between contexts (Hypothesis 8).
ship between affective empathy and (cyber)bullying, we were unable to
find an association on this component level, joining a few studies that 5.4. Covariates
have similarly found a lack of connection (e.g., Gano-Overway, 2013;
Kokkinos & Kipritsi, 2012). The narrow age span of our sample could be 5.4.1. Relationships with traditional and cyberbullying
an explanation for the nonsignificant relationships. For example, Ciucci Contrary to the existing literature (e.g., Kowalski et al., 2014), we
and Baroncelli (2013) found that affective empathy only predicted found no relationships between gender and age and self-reported in-
bullying in 6th graders but not in 8th graders. However, the non-rela- volvement in cyber- or traditional bullying after controlling for em-
tion might also be a consequence of the differential relationships be- pathy and all other covariates. One explanation for this finding could lie
tween various subcomponents and (cyber)bullying we observed in our in the nature of our sample, which represented a relatively narrow age
study. This could have masked the component level effect. Cognitive range and a slightly asymmetric gender distribution. Nevertheless, this
empathy, as expected, was negatively related to traditional and cy- should not have affected our main results since we focused on global
berbullying, with a stronger relation for traditional than for cyberbul- relationships on a variable and not an individual level. Furthermore,
lying. This is in line with our hypotheses (Hypothesis 2 and Hypothesis contrary to our assumptions (e.g., Greitemeyer & Mügge, 2014),
4). With respect to cognitive empathy, our results indicate that the gaming attitudes were not related to either cyber- or traditional bul-
ability to understand vicarious emotions is a protective factor against lying. We suspect that this nonsignificant result was caused by the fact
both cyber- and traditional bullying. This is in line with existing studies that we didn't control for the type of video games played. However,
(van Noorden et al., 2015). Additionally, the protective effect seemed since this was not part of our main questions, further studies should
significantly stronger in the traditional than in the cyber context. This investigate this relationship in greater detail.
finding underpins our assumptions as well as Runions' (2013) theore- In line with our assumptions, we found a positive relationship be-
tical considerations and indicates that eliciting cognitive empathy may tween social media use and cyberbullying in Model 2 investigating
be more difficult in the cyber context than in the face-to-face context. components of empathy. While traditional bullying was also positively
related to social media use in this model, this association disappeared in
5.2. Subcomponents of affective empathy Model 4 (investigating subcomponents of affective and cognitive em-
pathy). For cyberbullying, the positive relation remained in Model 4.
Although we hypothesized relationships between all subcomponents We were able to replicate existing findings regarding the assumed po-
of affective empathy and both forms of bullying (Hypothesis 6), we only sitive relation between migration background and traditional bullying
found a significant negative relationship between peripheral re- (e.g., Fandrem et al., 2009; Strohmeier et al., 2012). This finding re-
sponsivity and cyberbullying (Model 3 and Model 4) and a significant mained stable across Models 3 and 4. No relations were found between
positive relationship between emotion contagion and traditional but migration background and cyberbullying.
not cyberbullying (Model 4). For all other subcomponents of affective
empathy, there were no significant relationships with either cyber- or 5.4.2. Invariance of relationships between components and subcomponents
traditional bullying. Nevertheless, in line with our assumptions, the of empathy and both forms of bullying
negative relationship between peripheral responsivity and cyberbul- Regarding the components of empathy, the pattern of relationships
lying was significantly stronger than its non-significant relationship remained stable after considering and controlling for covariates
with traditional bullying (Hypothesis 9). That is, higher sensitivity to (Hypothesis 5). With respect to the subcomponents of cognitive and
cues in a remote social context (e.g., a person perceived as a character affective empathy (Hypothesis 11), including covariates revealed a
in a film) seems to be a protective factor for cyberbullying but not positive relationship between emotion contagion and traditional but
traditional bullying. Although the conclusions should be tentative, not cyberbullying. All other relationships were invariant. This finding
these results suggest the presence of contextual differences in in- indicates that emotion contagion as a precursor of affective empathy
formation processing and thus support the online disinhibition effect may be a unique risk factor for traditional bullying. More specifically, a
(Suler, 2004) and social presence theory (e.g., Short et al., 1976; lack of ability to make self-other distinctions may lead to more stressful
Gunawardena, 2004) by demonstrating the importance of the modality reactions to negative emotions among others, which could in turn lead
in which communication occurs. Furthermore, in line with our hy- to aggressive behavior and, if this process repeats, bullying. Our results
pothesis (Hypothesis 10), proximal responsivity was more strongly re- suggest that stressful reactions to others' emotions may be more evident
lated to traditional than cyberbullying. However, whether this differ- in the traditional than in the cyber context. Further research should
ence has practical implications remains questionable, because these address this assumption.
relationships were not statistically significant. The positive relationship
between emotion contagion and traditional bullying is discussed in 5.5. Limitations
Section 5.4.2.
Although empathy should be considered a precursor of (cyber)bullying
5.3. Subcomponents of cognitive empathy (Zych et al., 2018), our cross-sectional data does not allow for such a
causal interpretation. Longitudinal and experimental research is needed to
With respect to the subcomponents of cognitive empathy, we found substantiate our findings. Furthermore, in terms of methodological issues,
negative relationships between online simulation and both forms of our data relies on self-report measures, which may result in an under-
bullying. Moreover, we observed positive relations between perspective representation of reported bullying involvement due to social desirability
taking and both forms of bullying. This is in line with our hypothesis (Beran, Rinaldi, Bickham & Rick, 2012). Further research should use a
(Hypothesis 7). Our results indicate that the ability to understand an- mixed-method approach involving different forms of data collection such
other person's emotions intuitively may be insufficient to act as a pro- as peer nominations or ratings and information from other persons such as
tection factor against (cyber)bullying. On the contrary, our results parents or teachers to overcome this limitation. Moreover, our sample was

193
D. Graf, et al. Computers in Human Behavior 96 (2019) 186–195

characterized by a relatively narrow age range and a slightly asymmetric commonalities in empathy's role in bullying across contexts (face-to-
gender distribution. Further research should investigate the stability of our face, cyberspace). Although a large body of literature suggests that
findings across a wider age span and among different age groups. Future empathy plays an exclusively protective role in (cyber)bullying, we
research should also investigate whether there are differences in the pat- were able to reveal not only just protecting but also promoting func-
terns of relations in different subgroups (e.g., bully-victims, see Smith, tions of empathy in (cyber)bullying by examining relationships be-
1991; or reinforcers, see Salmivalli, 1999). Finally, we recommend that tween subcomponents of affective and cognitive empathy and cyber- as
future research apply a person-oriented approach and corresponding sta- well as traditional bullying. In our study, online simulation was nega-
tistical methods, such as latent class analysis or cluster analysis (von Eye & tively associated with both forms of bullying and peripheral re-
Spiel, 2010). In our study, we only applied a variable oriented approach. sponsivity was negatively related to cyberbullying; on the contrary,
perspective taking was positively related to both forms of bullying and
5.6. Theoretical contribution and practical implications emotion contagion was positively related to traditional bullying. Thus,
on the one hand, our study contributes to the discussion of how cog-
Empathy was repeatedly reported as a common predictor of both nitive empathy relates to (cyber)bullying. On the other hand, we de-
traditional and cyberbullying (e.g., Ang & Goh, 2010; Del Rey et al., monstrate differential roles of subcomponents of affective empathy in
2016; Jolliffe & Farrington, 2006; Steffgen et al., 2011). However, traditional and cyberbullying.
studies investigating the relationship between empathy and (cyber) Furthermore, by investigating differences in the strength of the re-
bullying mostly neglected empathy's multifaceted nature and assessed lationships between components and subcomponents of empathy and
the construct more broadly. The current study investigated the roles of cyber- as well as traditional bullying, we contribute to discussions re-
empathy (sub)components in relation to bullying, while at the same garding differences between cyber- and traditional bullying with re-
time considering the different contexts (face-to-face, cyberspace) in spect to empathy. On a component level, we found that the protective
which bullying occurs. This approach revealed differential relationships role of cognitive empathy seemed quantitatively stronger in traditional
between subcomponents of empathy, traditional and cyberbullying. than in cyberbullying. Further, peripheral responsivity was more
Moreover, quantitative differences in the strength of the associations strongly negatively related to cyberbullying and proximal responsivity
between (sub)components and (cyber)bullying occurred. Although was more strongly negatively related to traditional bullying.
there was some overlap with traditional bullying, independent features Our findings provide a basis for the development of environment-
were observable for cyberbullying with respect to empathy. specific prevention and intervention programs and contribute to current
While several limitations apply to the reported results, prevention discussions regarding the role of empathy in cyber- and traditional
and intervention programs tackling (cyber)bullying are advised to bullying.
consider the context (face-to-face, cyberspace) in which bullying ap-
pears. Based on our findings with respect to affective empathy, per- Conflicts of interest
ipheral responsivity may be seen as a protective factor against cyber-
bullying but not traditional bullying. Programs focusing primarily on There are no competing interests.
cyberbullying may benefit from focusing on tackling the online disin-
hibition effect (Suler, 2004). This may be achieved by targeting self- Funding
awareness, particularly concerning the effects of one's own aggressive
behavior on real human beings. Thus, fostering online awareness might This research did not receive any specific grants from funding
be helpful to counteract subconscious psychological processes asso- agencies in the public, commercial, or not-for-profit sectors.
ciated with the online disinhibition effect (Terry & Cain, 2016).
In contrast, emotion contagion might be a unique risk factor for References
traditional bullying. Training the ability to differentiate oneself from
others might help diminishing reactions of empathy-related stress and Anderson, C. A., & Bushman, B. J. (2002). Human aggression. Annual Review of Psychology,
thereby stress-related aggressive behavior. 53(1), 27–51. https://doi.org/10.1146/annurev.psych.53.100901.135231.
Anderson, C. A., Shibuya, A., Ihori, N., Swing, E. L., Bushman, B. J., Sakamoto, A., ...
Based on our results, cognitive empathy trainings might be seen as a Saleem, M. (2010). Violent video game effects on aggression, empathy, and prosocial
more universal approach to prevent both cyber- and traditional bul- behavior in Eastern and Western countries: A meta-analytic review. Psychological
lying, with a potentially larger effect in the context of traditional bul- Bulletin, 136(2), 151–173. https://doi.org/10.1037/a0018251.
Ang, R. P., & Goh, D. H. (2010). Cyberbullying among adolescents: The role of affective
lying. However, exclusively training perspective taking (i.e. the un- and cognitive empathy, and gender. Child Psychiatry and Human Development, 41(4),
derstanding of others' emotions) may be contraindicated. This is due to 387–397. https://doi.org/10.1007/s10578-010-0176-3.
the assumption that bullies might exploit their perspective taking- Antoniadou, N., & Kokkinos, C. M. (2018). Empathy in traditional and cyber bullying/
victimization involvement from early to middle adolescence: A cross sectional study.
ability to bully their victims more effectively. Instead, focusing on on-
Journal of Educational and Developmental Psychology, 8(1), 153. https://doi.org/10.
line simulation (i.e. the effort to adopt another person's perspective and 5539/jedp.v8n1p153.
to mentally represent another person's emotional state) might be a Baron-Cohen, S., Richler, J., Bisarya, D., Gurunathan, N., & Wheelwright, S. (2003). The
systemizing quotient: An investigation of adults with asperger syndrome or high-func-
promising approach to combat both traditional and cyberbullying.
tioning autism, and normal sex differences. Philosophical Transactions of the Royal Society B:
Further studies are advised to investigate other factors related to Biological Sciences, 358(1430), 361–374. https://doi.org/10.1098/rstb.2002.1206.
bullying while considering different bullying contexts. For example, Baron-Cohen, S., & Wheelwright, S. (2004). The empathy quotient: An investigation of
considering underlying risk factors such as sensation seeking and ag- adults with asperger syndrome or high functioning autism, and normal sex differ-
ences. Journal of Autism and Developmental Disorders, 34(2), 163–175. https://doi.
gression motivation might provide us further in-depth information on org/10.1023/b:jadd.0000022607.19833.00.
how cyberbullying differs from traditional bullying. Beran, T. N., Rinaldi, C., Bickham, D. S., & Rich, M. (2012). Evidence for the need to
support adolescents dealing with harassment and cyber-harassment: Prevalence,
progression, and impact. School Psychology International, 33(5), 562–576. https://doi.
5.7. Summary and conclusion org/10.1177/0143034312446976.
Best, P., Manktelow, R., & Taylor, B. (2014). Online communication, social media and
The aim of the present study was to investigate the role of empathy adolescent wellbeing: A systematic narrative review. Children and Youth Services
Review, 41, 27–36. https://doi.org/10.1016/j.childyouth.2014.03.001.
in (cyber)bullying by accounting for empathy's multifaceted nature and Bovaird, J. A., & Koziol, N. A. (2012). Measurement models for ordered-categorical in-
two contexts in which bullying appears (face-to-face, cyberspace). By dicators. In R. H. Hoyle (Ed.). Handbook of structural equation modeling (pp. 495–511).
examining not only components but also subcomponents of empathy New York, NY: Guilford Press.
Brown, T. A. (2015). Confirmatory factor analysis for applied research (2nd ed.). New York:
and their relations to cyber- as well as traditional bullying, our study Guilford Press.
provides insights into functional and quantitative differences and

194
D. Graf, et al. Computers in Human Behavior 96 (2019) 186–195

Casas, J. A., Del Rey, R., & Ortega-Ruiz, R. (2013). Bullying and cyberbullying: s10964-014-0135-6.
Convergent and divergent predictor variables. Computers in Human Behavior, 29(3), Olweus, D. (1993). Bullying at school: What we know and what we can do. Oxford:
580–587. https://doi.org/10.1016/j.chb.2012.11.015. Blackwell.
Ciucci, E., & Baroncelli, A. (2014). The emotional core of bullying: Further evidences of Olweus, D. (2012). Cyberbullying: An overrated phenomenon? European Journal of
the role of callous–unemotional traits and empathy. Personality and Individual Developmental Psychology, 9(5), 520–538. https://doi.org/10.1080/17405629.2012.
Differences, 67, 69–74. https://doi.org/10.1016/j.paid.2013.09.033. 682358.
Davis, M. H. (1983). Measuring individual differences in empathy: Evidence for a mul- Olweus, D. (2013). School bullying: Development and some important challenges. Annual
tidimensional approach. Journal of Personality and Social Psychology, 44(1), 113–126. Review of Clinical Psychology, 9(1), 751–780. https://doi.org/10.1146/annurev-
https://doi.org/10.1037/0022-3514.44.1.113. clinpsy-050212-185516.
Decety, J. (2011). Dissecting the neural mechanisms mediating empathy. Emotion Review, R Core Team. (2018). R: A language and environment for statistical computing. Vienna,
3(1), 92–108. https://doi.org/10.1177/1754073910374662. Austria: R Foundation for Statistical Computing.
Decety, J., & Jackson, P. L. (2004). The functional architecture of human empathy. Reniers, R. L. E. P., Corcoran, R., Drake, R., Shryane, N. M., & Völlm, B. A. (2011). The
Behavioral and Cognitive Neuroscience Reviews, 3(2), 71–100. https://doi.org/10. QCAE: A questionnaire of cognitive and affective empathy. Journal of Personality
1177/1534582304267187. Assessment, 93(1), 84–95. https://doi.org/10.1080/00223891.2010.528484.
Decety, J., & Jackson, P. L. (2006). A social-neuroscience perspective on empathy. Current Revelle, W. (2018). psych: Procedures for personality and psychological research. Evanston,
Directions in Psychological Science, 15(2), 54–58. https://doi.org/10.1111/j.0963- Illinois, USA: Northwestern University. https://CRAN.R-project.org/package=
7214.2006.00406.x. psychVersion=1.8.4.
Decety, J., & Lamm, C. (2006). Human empathy through the lens of social neuroscience. Runions, K. C. (2013). Toward a conceptual model of motive and self-control in cyber-
The Scientific World Journal, 6, 1146–1163. https://doi.org/10.1100/tsw.2006.221. aggression: Rage, revenge, reward, and recreation. Journal of Youth and Adolescence,
Decety, J., & Meyer, M. (2008). From emotion resonance to empathic understanding: A 42(5), 751–771. https://doi.org/10.1007/s10964-013-9936-2.
social developmental neuroscience account. Development and Psychopathology, Runions, K. C., & Bak, M. (2015). Online moral disengagement, cyberbullying, and cyber-
20(04), 1053–1080. https://doi.org/10.1017/S0954579408000503. aggression. Cyberpsychology, Behavior, and Social Networking, 18(7), 400–405. https://
Del Rey, R., Casas, J. A., Ortega-Ruiz, R., Schultze-Krumbholz, A., Scheithauer, H., Smith, doi.org/10.1089/cyber.2014.0670.
P., et al. (2015). Structural validation and cross-cultural robustness of the European Salmivalli, C. (1999). Participant role approach to school bullying: Implications for inter-
cyberbullying intervention Project questionnaire. Computers in Human Behavior, 50, ventions. Journal of Adolescence, 22(4), 453–459. https://doi.org/10.1006/jado.1999.
141–147. https://doi.org/10.1016/j.chb.2015.03.065. 0239.
Del Rey, R., Lazuras, L., Casas, J. A., Barkoukis, V., Ortega-Ruiz, R., & Tsorbatzoudis, H. Short, J., Williams, E., & Christie, B. (1976). The social Psychology of telecommunications.
(2016). Does empathy predict (cyber) bullying perpetration, and how do age, gender London: Wiley Press.
and nationality affect this relationship? Learning and Individual Differences, 45, Smith, P. K. (1991). The silent nightmare: Bullying and victimization in school peer
275–281. https://doi.org/10.1016/j.lindif.2015.11.021. groups. The Psychologist: Bulletin of the British Psychological Society, 4(6), 243–248.
Eisenberg, N., & Miller, P. A. (1987). The relation of empathy to prosocial and related be- Smith, P. K., & Ananiadou, K. (2003). The nature of school bullying and the effectiveness
haviors. Psychological Bulletin, 101(1), 91–119. https://doi.org/10.1037/0033-2909.101. of school-based interventions. Journal of Applied Psychoanalytic Studies, 5(2),
1.91. 189–209. https://doi.org/10.1023%2FA%3A1022991804210.
Enders, C. (2010). Applied missing data analysis. New York, NY: Guilford. Smith, P. K., Mahdavi, J., Carvalho, M., Fisher, S., Russell, S., & Tippett, N. (2008).
von Eye, A., & Spiel, C. (2010). Conducting person-oriented research. Zeitschrift für Cyberbullying: Its nature and impact in secondary school pupils. Journal of Child
Psychologie/Journal of Psychology, 218, 151–154. https://doi.org/10.1027/0044- Psychology and Psychiatry, 49(4), 376–385. https://doi.org/10.1111/j.1469-7610.
3409/a000024. 2007.01846.x.
Fandrem, H., Strohmeier, D., & Roland, E. (2009). Bullying and victimization among Steffgen, G., König, A., Pfetsch, J., & Melzer, A. (2011). Are cyberbullies less empathic?
native and immigrant adolescents in Norway: The role of proactive and reactive Adolescents' cyberbullying behavior and empathic responsiveness. Cyberpsychology,
aggressiveness. The Journal of Early Adolescence, 29(6), 898–923. https://doi.org/10. Behavior, and Social Networking, 14(11), 643–648. https://doi.org/10.1089/cyber.
1177/0272431609332935. 2010.0445.
Gano-Overway, L. A. (2013). Exploring the connections between caring and social be- Sticca, F., & Perren, S. (2013). Is cyberbullying worse than traditional bullying?
haviors in physical education. Research Quarterly for Exercise & Sport, 84(1), 104–114. Examining the differential roles of medium, publicity, and anonymity for the per-
https://doi.org/10.1080/02701367.2013.762322. ceived severity of bullying. Journal of Youth and Adolescence, 42(5), 739–750. https://
Georgi, E., Petermann, F., & Schipper, M. (2014). Are empathic abilities learnable? doi.org/10.1007/s10964-012-9867-3.
Implications for social neuroscientific research from psychometric assessments. Social Strohmeier, D., Fandrem, H., & Spiel, C. (2012). The need for peer acceptance and af-
Neuroscience, 9(1), 74–81. https://doi.org/10.1080/17470919.2013.855253. filiation as underlying motive for aggressive behaviour and bullying others among
Greitemeyer, T., & Mügge, D. O. (2014). Video games do affect social outcomes: A meta- immigrant youth living in Austria and Norway. Anales de Psicología/Annals of
analytic review of the effects of violent and prosocial video game play. Personality and Psychology, 28(3), 695–704. https://doi.org/10.6018/analesps.28.3.155991.
Social Psychology Bulletin, 40(5), 578–589. https://doi.org/10.1177/ Suler, J. (2004). The online disinhibition effect. CyberPsychology and Behavior, 7(3),
0146167213520459. 321–326. https://doi.org/10.1089/1094931041291295.
Gunawardena, C. N. (2004). Designing the social environment for online learning: The Sutton, J., Smith, P. K., & Swettenham, J. (1999). Social cognition and bullying: Social
role of social presence. In D. Murphy, R. Carr, J. Taylor, & T. Wong (Eds.). Distance inadequacy or skilled manipulation? British Journal of Developmental Psychology,
education and technology: Issues and practice (pp. 255–270). Hong Kong: Open 17(3), 435–450. https://doi.org/10.1348/026151099165384.
University of Hong Kong Press. Sutton, J., Smith, P. K., & Swettenham, J. (2001). Bullying and ‘theory of mind’: A critique
Hogan, R. (1969). Development of an empathy scale. Journal of Consulting and Clinical of the ‘social skills deficit’ view of anti-social behaviour. Social Development, 8(1),
Psychology, 33(3), 307–316. https://doi.org/10.1037/h0027580. 117–127. https://doi.org/10.1111/1467-9507.00083.
Jolliffe, D., & Farrington, D. P. (2006). Examining the relationship between low empathy and Terry, C., & Cain, J. (2016). The emerging issue of digital empathy. American Journal of
bullying. Aggressive Behavior, 32(6), 540–550. https://doi.org/10.1002/ab.20154. Pharmaceutical Education, 80(4), 1–4. https://doi.org/10.5688/ajpe80458.
Juvonen, J., & Gross, E. F. (2008). Extending the school grounds?-bullying experiences in Tokunaga, R. S. (2010). Following you home from school: A critical review and synthesis
cyberspace. Journal of School Health, 78(9), 496–505. https://doi.org/10.1111/j. of research on cyberbullying victimization. Computers in Human Behavior, 26(3),
1746-1561.2008.00335.x. 277–287. https://doi.org/10.1016/j.chb.2009.11.014.
Kline, R. B. (2015). Principles and practice of structural equation modeling (4th ed.). Boca Vachon, D. D., Lynam, D. R., & Johnson, J. A. (2014). The (non)relation between empathy
Raton: Chapman & Hall/CRC Press. and aggression: Surprising results from a meta-analysis. Psychological Bulletin, 140(3),
Kokkinos, C. M., & Kipritsi, E. (2012). The relationship between bullying, victimization, 751–773. https://doi.org/10.1037/a0035236.
trait emotional intelligence, self-efficacy and empathy among preadolescents. Social Vivolo-Kantor, A. M., Martell, B. N., Holland, K. M., & Westby, R. (2014). A systematic
Psychology of Education, 15(1), 41–58. https://doi.org/10.1007/s11218-011-9168-9. review and content analysis of bullying and cyber-bullying measurement strategies.
Kowalski, R. M., Giumetti, G. W., Schroeder, A. N., & Lattanner, M. R. (2014). Bullying in the Aggression and Violent Behavior, 19(4), 423–434. https://doi.org/10.1016/j.avb.2014.
digital age: A critical review and meta-analysis of cyberbullying research among youth. 06.008.
Psychological Bulletin, 140(4), 1073–1137. https://doi.org/10.1037/a0035618. Walter, H. (2012). Social cognitive neuroscience of empathy: Concepts, circuits, and
Kowalski, R. M., & Limber, S. P. (2007). Electronic bullying among middle school stu- genes. Emotion Review, 4(1), 9–17. https://doi.org/10.1177/1754073911421379.
dents. Journal of Adolescent Health, 41(6), 22–30. https://doi.org/10.1016/j. Walther, J. B. (2011). Theories of computer- mediated communication and interpersonal
jadohealth.2007.08.017. relations. In M. L. Knapp, & J. A. Daly (Eds.). The handbook of interpersonal commu-
Little, T. D., Slegers, D. W., & Card, N. (2006). A non-arbitrary method of identifying and nication (pp. 443–479). (4th ed.). Thousand Oaks, CA: Sage.
scaling latent variables in SEM and MACS models. Structural Equation Modeling, 13, Walther, J. B., & Burgoon, J. K. (1992). Relational communication in computer-mediated
59–72. https://doi.org/10.1207/s15328007sem1301_3. interaction. Human Communication Research, 19(1), 50–88. https://doi.org/10.1111/
Mishna, F., Saini, M., & Solomon, S. (2009). Ongoing and online: Children and youth's j.1468-2958.1992.tb00295.x.
perceptions of cyber bullying. Children and Youth Services Review, 31(12), 1222–1228. Zaki, J., & Ochsner, K. N. (2012). The neuroscience of empathy: Progress, pitfalls and
https://doi.org/10.1016/j.childyouth.2009.05.004. promise. Nature Neuroscience, 15(5), 675–680. https://doi.org/10.1038/nn.3085.
Muthén, L. K., & Muthén, B. O. (1998–2018). Mplus user's guide (8th ed.). Los Angeles, CA: Zych, I., Baldry, A. C., Farrington, D. P., & Llorent, V. J. (2018). Are children involved in
Muthén & Muthén. cyberbullying low on empathy? A systematic review and meta-analysis of research on
van Noorden, T. H. J., Haselager, G. J. T., Cillessen, A. H. N., & Bukowski, W. M. (2015). empathy versus different cyberbullying roles. Aggression and Violent Behavior. in press
Empathy and involvement in bullying in children and adolescents: A systematic re- https://doi.org/10.1016/j.avb.2018.03.004.
view. Journal of Youth and Adolescence, 44(3), 637–657. https://doi.org/10.1007/

195

Das könnte Ihnen auch gefallen