Longitudinal associations between psychopathic traits and social support with cyberbullying and cyber-victimization

ABSTRACT

The objective of the study was to examine how distinct psychopathic traits, including callous-unemotional (CU) traits, impulsivity and narcissism, as well as family, school and friend social support are longitudinally associated with cyberbullying and cyber-victimization. Associations were investigated by employing a four-year longitudinal design in a large sample of Greek-Cypriot adolescents. Findings suggested that cyber-bullying and cyber-victimization decreased from early to middle adolescence. Additionally, the longitudinal model suggested that both CU traits and narcissism, assessed in grade 7, were associated with the onset of cyber-bullying and cyber-victimization in grade 8, but not with change over time. Impulsivity did not predict cyber-bullying or cyber-victimization. Finally, none of the environmental factors were associated with the onset and growth of cyber-bullying, although family support predicted girls’ initial levels and change over time in cyber-victimization. Current findings have major theoretical and practical implications for the development of cyber-bullying and victimization.

KEYWORDS:

• Cyber-bullying
• cyber-victimization
• narcissism
• impulsivity
• callous-unemotional traits

Introduction

With the frequent use of internet and communication media as well as social network sites by children and adolescents, cyberbullying has become a major youth problem worldwide (Fanti & Zacharaki, 2021; Kowalski et al., 2014). Cyber-bullying is defined as an aggressive, intentional act carried out by an individual or a group of individuals with the use of social media and other forms of electronic contact (Smith et al., 2008). This action is repeated across time and the victim cannot easily defend him or herself (Strom & Strom, 2005). Specifically, cyber-bullies may repeatedly harass, tease, disrespect, spread rumours, or exclude from social activities their peers with the use of instant messaging, chat rooms, email, and text messages through cell phones, computers and social networking sites. Thus, similar to bullying, cyber-bullying is intentional, and the action is repeated over time.

Although the majority of prior work focuses on school bullying, cyber-bullying has been found to be similarly stressful and harmful to its victims (Kowalski et al., 2014). Additional work suggested that cyber-victimization might even be associated with higher social and psychological maladjustment compared to traditional forms of victimization (Van Geel et al., 2014). Importantly, cyber-bullying influences one in five adolescents, with some studies reporting that up to 50% of adolescents might engage or be the recipients of cyber-bullying (Kowalski et al., 2014; Modecki et al., 2014). The fact that cyber-bullying is more prominent during adolescence, which is a vulnerable developmental period associated with lower emotion regulation as well as increased externalizing and internalizing problems (Fanti, 2013; Steinberg et al., 2018), can bring additional strains that adolescents might find extremely difficult to handle. As such, cyber-bullying is regarded as a public health concern (Nixon, 2014).

With the aim of informing individualized interventions, the current study applies the ecological model of development as a theoretical framework to understand cyber-bullying and cyber-victimization (Bronfenbrenner, 1994). Following the ecological model, the individual’s personality traits and the social contexts in which the individual is embedded are considered. On the individual level, we explore possible associations with psychopathic personality traits, including callous-unemotional (CU) traits, narcissism, and impulsivity (Fanti et al., 2018). At the environmental level, three protective variables in the child’s proximal environment, including family, school and friend social support, are taken into account.

Psychopathic personality traits

Psychopathic personality is a multidimensional syndrome consisting of a constellation of co-occurring interpersonal, affective, and behavioural traits. Research during adolescence suggests that callous-unemotional traits (i.e., lack of remorse or guilt; lack of concern for others’ feelings; and shallow or deficient emotions), narcissism (i.e., egocentricity, manipulation, grandiosity, and superficial charm), and impulsivity (irresponsibility, disinhibition and need for stimulation) are associated with severe child and adolescent antisocial behaviour, including aggression, conduct problems, bullying, and delinquency (see Fanti, 2018; Fanti et al., 2018; Kimonis et al., 2015; Salekin, 2018).

Youth high on impulsivity are characterized by a pervasive lack of behavioural restraint, which increases risk for aggressive tendencies with limited regard for long term consequences (Salekin, 2018). Although impulsivity is associated with traditional bullying, the association with cyberbullying is not consistent. For example, Liu et al. (2021) and Fanti et al. (2012) found no association between impulsivity and cyber-bullying, whereas López-Larrañaga and Orue (2019) found that impulsivity longitudinally predicted cyber-bullying. Thus, additional longitudinal studies are needed to examine potential associations between impulsivity and cyberbullying. In terms of victimization, findings suggest that impulsivity predicts both traditional and online victimization (Baroncelli et al., 2022; Fanti & Kimonis, 2012). Impulsive youth, who are characterized by impaired emotion regulation and negative affect, might be at increased risk for social rejection in both school and online contexts, increasing their chances of being victimized.

In contrast to impulsive adolescents, youth high on narcissism might choose to engage in cyber-bullying mainly as an effort to gain high social status or popularity as well as a dominant position in the peer network (Fanti & Henrich, 2015; van Geel et al., 2017). Although some contradicting findings exist (Fanti et al., 2012; Goodboy & Martin, 2015), several studies suggested that narcissism is one of the most important predictors of cyber-bullying (Fan et al., 2019; López-Larrañaga & Orue, 2019; van Geel et al., 2017). In terms of the mechanisms behind this association, cyber-bullying was found to enhance feelings of superiority and self-worth as well as the threaten ego characterizing individuals with narcissistic personality traits (Fan et al., 2019). Additionally, Fan et al. (2019) found that narcissism might also play an important role in cyber-victimization. However, this finding might be explained by the higher likelihood of narcissistic individuals to perceive themselves as victims, which might be associated with their higher reports of cyber-victimization (Fanti & Henrich, 2015; Fanti & Kimonis, 2012; McCullough et al., 2003).

CU traits have been associated with both bullying and cyber-bullying in several studies, suggesting that the lack of concern over others’ feelings might characterize bullies (Antoniadou et al., 2019; Baroncelli et al., 2022; Baumann et al., 2022; Fanti, 2013). Indeed, individuals that lack empathy and show disregard for others’ feelings might be more likely to be involved in repeated bullying against a victim that cannot defend his or herself (Ciucci & Baroncelli, 2014; Wright et al., 2019). In addition, the distress of victims might be less inhibitory for youth high on CU traits (Fanti et al., 2012). An interesting finding was that adolescents engaging in cyberbullying tend to experience less empathy for their victims, compared to those who engage in traditional forms of bullying (Steffgen et al., 2011). Thus, low empathy and callous or unemotional traits might be important individual characteristics of cyber-bullies. Moreover, CU traits were uniquely and longitudinally associated with cyberbullying even after accounting for conduct problems as well as school bullying and victimization, suggesting that these traits might be one of the most important predictors of online bullying (Baroncelli et al., 2022; Fanti, 2013; Fanti et al., 2012; Orue & Calvete, 2019). CU traits were also associated with cyber-victimization cross-sectionally and longitudinally (Baroncelli et al., 2022; Baumann et al., 2022; Fanti et al., 2012). However, the majority of prior work suggests that CU traits are mainly associated with cyber-bullying but not victimization (e.g., Fanti & Kimonis, 2012; Fanti et al., 2012).

Family, school, and peer support

To understand complex social phenomena like cyberbullying, it is important to also consider environmental factors that protect the developing person from engaging in them (Kowalski et al., 2019). Such protective factors can be often found in the proximal environment of a person, including the family, the school, and the peer systems. Overall, past literature shows that perceived support from these systems tends to be associated with lower risk for involvement in cyberbullying and cybervictimization (Fanti et al., 2012; Williams & Guerra, 2007; also see; Kowalski et al., 2014 for a meta-analysis).

Literature on traditional forms of bullying has shown that perceiving the school personnel as supportive might protect victims of bullying because it enables them to seek help (Eliot et al., 2010). While there is some evidence suggesting that this protective effect might extend to cybervictimization (Ortega-Barton et al., 2016), other studies fail to find an association between school support and cybervictimization (Álvarez-García et al., 2015; Fanti et al., 2012). Furthermore, there is evidence suggesting that school support might be a protective factor for cyberbullying perpetration (Baldry et al., 2015), but more studies are needed to establish this finding. Perceived peer support has also been suggested as a potential protective factor for both cybervictimization (Baldry et al., 2015; Fridh et al., 2015; Kowalski et al., 2014) and cyberbullying perpetration (Arató et al., 2022; Williams & Guerra, 2007). However, a recent meta-analysis concluded that the quality of relationship with peers is not a protective factor neither for perpetration nor cyber-victimization (Marciano et al., 2020).

In addition to school climate and peer social support, prior work also focused on the protective effects of family support, with several studies suggesting that family support serves as a protective factor for both cyber-bullying and victimization (e.g., Arató et al., 2022; Ates et al., 2018; Buelga et al., 2017; Fanti et al., 2012). In fact, family support was the only environmental factor that was associated longitudinally with cyber-bullying and cyber-victimization (Fanti et al., 2012). Thus, compared to peer and school social context, family support might be a more important protective factor for cyber-bullying and cyber-victimization.

Current study

Following the ecological model, the aim of the present study is to examine the unique longitudinal associations between psychopathic traits and distinct environmental experiences with the development of cyber-bullying and cyber-victimization. By accounting for the covariance between these factors, we will investigate whether distinct social contexts function independently from individual characteristics in influencing cyber forms of bullying and victimization. Importantly, such a longitudinal investigation can clarify prior contradicting findings and point to important risk and protective factors that might be associated with cyber-bullying and cyber-victimization. Building on prior work (e.g., Fanti et al., 2012), we also control for delinquency in the longitudinal model in order to identify the unique effects of individual and contextual variables above and beyond levels of behavioural problem. Based on the literature review, we expect that CU traits might be the strongest predictor of the development of cyber-bullying, and that family support might be a protective factor for both cyber-bullying and victimization.

Finally, gender is considered as a potential moderator of the association between cyber-bullying and victimization with individual and contextual predictors. Boys are more likely to engage in cyber-bullying (Fanti et al., 2012; Hoareau et al., 2019), although boys and girls are equally likely to be cyber-victimized (Calvete et al., 2010). In terms of associations between psychopathy and cyber-bullying, prior work found that gender did not moderate the association between psychopathic traits and cyber-bullying (Hoareau et al., 2019; Wright et al., 2020). Similarly, Charalampous et al. (2021) found that high levels of psychopathic traits were associated with both boys and girls cyber-bullying behaviours; however, psychopathy was mainly associated with boys’ cyber-victimization. Thus, it might be that boys with psychopathic traits are more likely to be cyber-victimized, which is not the case for cyber-bullying. Regarding interactions effects between gender and the three environmental factors under investigation, prior work suggested that family, peer, and school social support predicted similarly boys and girls cyber-bullying and cyber-victimization (Fanti et al., 2012).

Method

Participants and procedure

The sample consisted of Greek Cypriot adolescents, equally divided into boys (49.9%) and girls (50.1%), who completed the study’s measures longitudinally. Following approval of the study by the Cyprus Ministry of Education, 13 middle schools in three school districts (Lefkosia, Larnaca, Lemeso) in Cyprus were randomly selected. After approval of the study by the school boards, students were given an informed consent form for their parents to sign, and only students with parental consent were permitted to participate in the study. In the classroom, students were informed about the study and were also informed about their rights as participants. Questionnaires were administered longitudinally by trained research assistants at grade 7 (Time 1 n = 1451; M age = 12.12, SD = .55), grade 8 (Time 2 n = 1373; M age = 13.05, SD = .58), the beginning of grade 9 (Time 3 n = 1268; M age = 13.81, SD = .64), and at the end of grade 9 (Time 4 n = 1224; M age = 14.45, SD = .65). Attrition was due to an inability to contact students who had moved away or transferred to a different school. Attrition analysis did not reveal any differences on demographic or main variables (i.e., psychopathic traits and cyber-bullying) between adolescents participating longitudinally and adolescents participating in only one assessment. Further, there was not differential attrition based on sex. The sample was diverse in terms of parental educational levels (20.60% below high school education, 45.95% with a high school education, and 33.45% with a university degree) and parental marital status (7.2% came from single parent families).

Measures

Psychopathic traits (Time 1)

CU traits were measured with the Inventory of Callous-Unemotional traits (ICU; Frick, 2004), which is a 24-item self-report scale. The items of the ICU were derived from four of the original APSD items that more consistently loaded on the CU subfactor, and for each item six additional items were created to provide a more detailed assessment of CU traits (Frick, 2004). Narcissism (seven-items) and impulsivity (five-items) were measured with the Antisocial Process Screening Device-Youth report (APSD; Frick & Hare, 2001). All items were placed on a four-point scale (from 0 = ‘not at all true’ to 3 = ‘definitely true’). The items measuring CU traits (α = .80; e.g., ‘I do not show my emotions to others’), narcissism (α = .70; e.g., ‘I act charming or nice to get things I want’), and impulsivity (α = .73; e.g., ‘I do not plan ahead or leave things until the last moment’) formed internally consistent scales. Previous research has provided evidence for the validity of the self-reported versions of the ICU and APSD in community and high-risk samples in Cyprus, Germany and U.S (Fanti et al., 2009; Kimonis et al., 2015).

Supportive social relations (Time 1)

The Multidimensional Scale of Perceived Social Support (MSPSS; Zimet, Dahlem, Zimet, & Farley, 1988) was used to assess supportive relationships within three contexts: family (a = .82; e.g., ‘“I get the emotional support I need from my family”’), friend (a = .80; e.g., ‘“I can count on my friend when things go wrong”’), and school (a = .87; e.g., ‘“The staff at my school provides me the support and encouragement that I need”’). The participants respondent on a 4-point scale (from 0 Not at all true to 3 Definitely true). Prior work has provided evidence that the MSPSS is a valid and reliable measure of perceived social support during adolescence (Canty-Mitchell & Zimet, 2000).

Cyber-bullying and cyber-victimization (Times 2–4)

The Student Survey of Bullying Behavior-Revised (SSBB-R; Varjas, Meyers, & Hunt, 2006) was administered at Times 2, 3 and 4 to measure cyber-bullying and cyber-victimization. Participants indicated whether they had engaged in different types of bullying or how often different types of victimization happened to them on an ordinal scale of: never, once or twice a year, monthly, weekly, or daily. The SSBB-R includes 4 items assessing cyber-bullying (α = .86) and 4 items assessing cyber-victimization (α = .91). To assess cyber-bullying and cyber-victimization participants were asked how often they sent or receive a threatening or harassing (1) email (e.g., school email account), (2) instant message (e.g., Viber), (3) message in a chat room or social networking sites (e.g., Facebook, Instagram), and (4) Short Text Messages (SMS). Previous research using the SSBB-R successfully measured cyber-bullying and cyber-victimization in community samples of adolescents in Cyprus and the U.S. (Fanti et al., 2009; Varjas et al., 2006).

Plan of Analysis

The current study aimed to investigate the longitudinal associations between psychopathic traits and environmental protective factors with cyber-bullying and cyber-victimization. The model also includes cross-sectional correlations between all the predictors to test for unique effects of each variable after accounting for their covariance. Analyses proceeded in three stages. In the first stage, latent growth curve modelling analysis was conducted using the Mplus 8 software package (Muthén & Muthén, 2010) to investigate the trajectories of change for cyber-bullying and cyber-victimization from Time 2 to Time 4, as well as to identify the intercept (initial levels) and linear slope (change over time) for these variables. This type of growth model uses a polynomial function to model the relationship between the behaviour under investigation and assessment time (Muthén & Muthén, 2010). The analysis is based on a linear growth model, since 3 times of measurement were available. In the second stage, predictors were added in the model for cyber-bullying and cyber-victimization separately. In the third stage, multiple group models were computed to investigate potential moderating effects of sex, following Little’s (1997) statistical guidelines. With the use of the chi-square difference test, equality of the longitudinal structural paths was examined by comparing a model in which associations were constrained to be equal across groups to a model in which these associations were freely estimated. If the chi-square change test was significant, it was concluded that there were cross-group differences in time specific and over time associations. All analyses were conducted in Mplus 8 (Muthén & Muthén, 2010).

For all the analyses, three standard fit indexes were used to evaluate model fit: The Root Mean-square Error of Approximation (RMSEA), Standardized Root Mean Residual (SRMR), and the Comparative Fit Index (CFI). Cut-off values below .06 for RMSEA, below .08 for SRMR, and above .95 for CFI are considered a good fit (Bollen, 1989; Hu & Bentler, 1999). The Full Information Maximum Likelihood Estimator in Mplus 8 was utilized for all the analyses. This type of estimation accommodates even large proportions of missing data by estimating the full model using all the available information from all participants.

Results

Descriptive statistics

The means and standard deviations for the study’s variables are displayed in Table 1. Cross-sectional zero-order correlations suggested that psychopathic traits and environmental factors were significantly inter-correlated and the same was true for cyber-bullying and cyber-victimization. Further, psychopathic traits were longitudinally associated with cyber-bullying and cyber-victimization. School support was only correlated with cyber-bullying at time 4, whereas peer support was associated with cyber-victimization at time 3 and cyber-bullying at time 4. Family support was correlated with both cyber-bullying and victimization across all time points except for cyber-bullying at time 3. Thus, zero-order correlations suggest more consistent associations between psychopathic traits with cyber forms of bullying and victimization compared to environmental factors.

Table 1. Descriptive statistics and correlations among the main study variables (N = 1416).

	CU traits	NAR	IMP	School support	Peer support	Family support	Del.	CB T2	CV T2	CB T3	CV T3	CB T4	CV T4
NAR	.29**
IMP	.33**	.62**
School support	−.23**	−.12**	−.24**
Peer support	−.29**	−.10**	−.10**	.39**
Family support	−.33**	−.25**	−.25**	.39**	.35**
Delinquency	.39**	.48**	.49**	−.21**	−.13**	−.33**
CB (T2)	.21**	.16**	.12**	−.02	−.05	−.10**	.14**
CV (T2)	.20**	.18**	.15**	−.03	−.06	−.15**	.19**	.69**
CB (T3)	.29**	.19**	.17**	−.05	−.06	−.05	.16**	.25**	.21**
CV (T3)	.21**	.16**	.14**	−.04	−.12**	−.12**	.16**	.24**	.35**	.66**
CB (T4)	.26**	.15**	.14**	−.06*	−.08*	−.09**	.21**	.22**	.19**	.29**	.18**
CV (T4)	.19**	.15**	.14**	−.04	−.05	−.09**	.16**	.17**	.21**	.21**	.28**	.71**
Descriptives
Mean (SD)	21.24 (7.98)	4.92 (2.94)	4.43 (2.36)	5.59 (3.15)	7.75 (2.77)	9.19 (2.68)	3.15 (2.83)	0.68 (1.78)	0.79 (1.74)	.52 (1.94)	.66 (1.92)	0.50 (1.80)	0.55 (1.53)

Note. **p < .01; *p < .05.CU = callous-unemotional; NAR = narcissism; IMP = impulsivity; CB = cyber-bullying; CV = cyber-victimization; T = time.

Structural equation Model for cyber-bullying

To investigate the trajectories of change, separate linear growth curve models were estimated for cyber-bullying and cyber-victimization. The linear growth model for cyber-bullying fit the data well, χ²_{(1, N = 1571)} = 2.79, p = .10, RMSEA = .034 (RMSEA CI: 001|.084), SRMR = .013, CFI = .991. The unstandardized intercept (i = .76, SE = .07, p < .001) and linear slope (s = −.08, SE = .03, p < .05) were significant, suggesting that cyber-bullying decreased over time. The addition of the predictor variables as shown in Figure 1a also resulted in a model that fit the data well, χ²_{(8, N = 1571)} = 12.41, p = .13, RMSEA = .02 (RMSEA CI: 01|.04), SRMR = .01, CFI = .99. Τhe correlations between the predictors are shown in Table 1. Figure 1a only shows the significant associations. CU traits and narcissism were positively related to the cyber-bullying initial value (intercept), but not with the slope. None of the other associations were significant. However, a strong association was identified between the intercept and the slope, suggesting that those with high initial levels of cyber-bullying were less likely to show decreases over time.

Figure 1. Longitudinal model with cyber-bullying as outcome.

Structural equation Model for cyber-victimization

The linear growth model for cyber-victimization also fit the data well, χ²_{(1, N = 1571)} = .08, p = .78, RMSEA = .01 (RMSEA CI: 001|.044), SRMR = .002, CFI = .968. The unstandardized intercept (i = .91, SE = −.11, p < .001) and linear slope (s = −.11, SE = .030, p < .001) were significant, suggesting that participants exhibited a decrease in cyber-victimization over time (Figure 1b). The SEM model with predictors fit the data well, χ²_{(8, N = 1571)} = 8.25, p = .41, RMSEA = .01 (RMSEA CI: .01|.03), SRMR = .01, CFI = .99. CU traits and narcissism increased the onset of cyber-victimization, whereas family social support was negatively associated with the intercept. Only family support was associated with the slope of cyber-victimization, indicating that adolescents experiencing high family support were more likely to show decreases in cyber- victimization over time. The correlation between the intercept and the slope suggested that those scoring initially high on cyber-victimization were less likely to show decreases in cyber-victimization.

Figure 1. Longitudinal model with cyber-victimization as outcome.

Multi-group path model

To investigate whether the longitudinal associations among the variables under investigation differed between boys and girls, a structural model that constrained the regression paths to be invariant was compared to a structural model with structural paths freely estimated with the use of the chi-square difference test. The difference between the models was only significant for the cyber-victimization growth model. Specifically, the unconstrained model fit the data better than the constrained model, with paths constrained to be equal, Δχ²_{(13, n = 1720)} = 73.05, p < .001, suggesting group differences in these associations. The difference in models was accounted for by two associations. Specifically, family support was negatively associated with girls (β = −.18, SE = .05, p < .01) but not boys (β = −.04, SE = .10, p = .71) initial levels of cyber-victimization, and similarly with girls (β = −.19, SE = .06, p < .01) but not boys (β = .05, SE = .10, p = .64) change over time in cyber-victimization.

Discussion

The present study investigates associations between distinct psychopathic traits and social support with cyberbullying and cybervictimization in adolescence. Findings contribute significantly to the existing literature by testing longitudinal associations over four waves of measurement. The moderating role of gender was also explored. Our longitudinal model suggests that both CU traits and narcissism, but not impulsivity, assessed in grade 7, were associated with the onset of cyber-bullying and cyber-victimization in grade 8, but not with decreases over time identified for both cyber-bullying and cyber-victimization. Despite significant correlations across time, delinquency did not predict cyber-bullying or cyber-victimization, after accounting for psychopathic traits. Finally, none of the environmental factors were associated with the onset and growth of cyber-bullying, although family support predicted girls’ initial levels and change over time in cyber-victimization.

Psychopathic traits as risk factors for future cyberbullying

Our findings confirm prior studies which identified CU traits as a unique predictor for cyberbullying even after controlling for other psychopathic traits (Baroncelli et al., 2022; Fanti et al., 2012; Orue & Calvete, 2019). Expanding this line of work, we provide novel evidence that those effects might be more important for the onset of cyber-bullying, but not the continuation of these behaviours. In fact, adolescents that start engaging in cyber-bullying are highly likely to continue doing so from early to middle adolescence. Furthermore, our findings strengthen the argument that the underlying mechanisms driving cyberbullying might be related to unique characteristics of the CU dimension. One possible underlying mechanism is related to impairments in the prosocial emotions that typically inhibit individuals from causing distress to others (i.e., empathy and guilt; Ciucci & Baroncelli, 2014; Fang et al., 2022; Steffgen et al., 2011). Thus, the lack of empathy and guilt, characterizing those high on CU traits, might drive cyber-bullying behaviours. Overall, our findings indicate that children with CU traits represent an important target group for cyberbullying interventions. For example, increasing the salience of victim’s distress and sadness in perpetrators’ perception could be effective in reducing cyber-bullying.

Similarly, narcissism was positively associated with the onset of cyberbullying after accounting for prior levels of psychopathic traits. The specific association with the onset of cyber-bullying during early adolescence could be attributed to developmental changes in self-esteem, which has been hypothesized to be an explanatory mechanism regarding this association (Fan et al., 2019; O’Moore & Kirkham, 2001). For instance, Birkeland et al. (2012) found that self-esteem tends to increase during adolescence. Therefore, in early adolescence, individuals with higher levels of narcissism may be more likely to engage in cyberbullying as a way to boost their self-esteem. However, as self-esteem continues to increase in middle adolescence, the positive association between narcissism and cyberbullying may weaken, which might explain the lower likelihood of narcissism to explain change over time in cyber-bullying. Alternatively, the weakened effect of narcissism on cyberbullying in middle adolescence may be associated with contextual rather than intra-individual changes. Children and adolescents with narcissistic traits use bullying as a means to gain social status among their peers. However, as adolescents grow older and participate in larger and more diverse peer groups, they may need to start employing alternative means of gaining popularity (Sentse et al., 2015).

In contrast to CU traits and narcissism, impulsivity was not associated with future cyberbullying. Our findings add to prior work which suggests that after accounting for other psychopathic traits, impulsivity does not uniquely predict cyberbullying (Baroncelli et al., 2022; Fanti et al., 2012; Liu et al., 2021). Comparing our findings with meta-analytic evidence suggesting a robust association between impulsivity and traditional bullying (van Geel et al., 2017), we could hypothesize that the underlying mechanisms driving traditional bullying in children with impulsive traits do not necessarily apply to cyberbullying. For example, it has been proposed that the association between impulsivity and bullying is explained via the lack of inhibition (van Geel et al., 2017). However, it is possible that this effect is weaker in cyberbullying because the majority of children tend to be less inhibited in the online world (online disinhibition effect; Suler, 2004). Furthermore, it has been suggested that children high on impulsivity might bully others in response to perceived provocation (van Geel et al., 2017). However, the ‘provocative’ behaviours of victims (see children’s own definitions by Guerin & Hennesy, 2002) usually happen in a group setting in the school context (Griffin & Gross, 2004), and thus they might not apply to cyberbullying.

Psychopathic traits as a risk factor for future cybervictimization

In comparison to cyberbullying perpetration, significantly fewer studies explored the relation of distinct psychopathic traits on cybervictimization. A few studies focused only on one dimension of psychopathic traits (e.g., Baumann et al., 2022; Fan et al., 2019), while other studies explored all three dimensions but with cross-sectional data (e.g., Antoniadou et al., 2019; Baroncelli et al., 2022). To the best of our knowledge, only Fanti et al. (2012) had explored distinct longitudinal effects on future cybervictimization. However, they did not find any significant longitudinal effects after accounting for the covariance between psychopathic traits. Therefore, the identified effects of narcissism and CU traits on future cybervictimization is a novelty of the current study. The increased likelihood of those high on CU traits to be cyber-victimized might be related to associated impairments in socio-affiliative processes, which increase youths vulnerability to be the recipients of bullying behaviours (Miron et al., 2020). According to prior work, the effect of narcissism on cyber-victimization could be attributed either to low self-esteem (Fan et al., 2019) or to over-reporting of victimization (Fanti & Henrich, 2015). Importantly, this is the first study suggesting that adolescents scoring high on narcissism and CU traits are highly likely to be both the perpetrators but also the victims of cyber forms of bullying.

Family, school, and peer support

Consistent with prior research (Fanti et al., 2012), out of the support systems that can be typically found in a child’s proximal environment, family appears to be the most important context for protecting from cybervictimization. Furthermore, two important contributions are introduced in the current study. Firstly, we show that perceived family support protects both from the initial levels of cyber-victimization and from change over time. Secondly, our data suggest that this protective effect is more prominent for girls. One possible explanation offered by Shaheen et al. (2019) is that male adolescents need more support from family to be protected from victimization. This might be related to gender-based variations in family socialization processes which result in boys experiencing family support differently. Nevertheless, for the interpretation of such findings cultural nuances should also be considered. Moreover, and in contrast to prior work (Fanti et al., 2012), family support did not protect from cyberbullying perpetration. Therefore, our data indicate that perpetration is more strongly affected by individual factors, such as CU traits and Narcissism.

In contrast to family support, school support does not appear to protect from cyberbullying or cybervictimization. Due to its occurrence outside the school context, it is possible that teachers and school personnel may pay less attention to cyber-bullying incidents. Therefore, this null finding underscores the importance of educating students and teachers about the potential impact of cyberbullying, as well as ways for the school to protect youth from these types of behaviours. Further, it might be that other variables within the school community, such as school climate and safety, might be more important for cyberbullying (e.g., Kowalski et al., 2019).

Additionally, we found that peer support does not protect from cyberbullying or cybervictimization, contradicting the series of other studies that report such associations (Arató et al., 2022; Baldry et al., 2015; Fridh et al., 2015; Kowalski et al., 2014; Williams & Guerra, 2007). While peer support showed a small negative correlation with future cybervictimization and cyberbullying in Times 3 and 4, respectively, the effect did not reach significance in our full model. Therefore, it is possible that family support might be a more important protective factor than peer support. Another angle to consider, is that we fail to find a significant association because we account for psychopathic traits. Therefore, it could be that the potentially protective effect of peer support does not extend to children with these traits because their relationships with peers might already be impaired (Haas et al., 2018).

Strengths & limitations

The key strength of the current study is the use of four waves of measurement, which enabled the investigation of how prior levels of psychopathic traits and social support experiences influenced the onset and development of cyber-bullying and cyber-victimization, increasing the validity of our findings. Further, we had a large sample size and relatively low attrition rate. Finally, testing the moderating role of gender is also an important strength as it revealed significant differences among boys and girls. At the same time, current findings should be considered in light of few limitations, such as the reliance on self-reports from adolescents.

Conclusions & future directions

In conclusion, the present study provides novel evidence on the longitudinal associations between psychopathic traits with cyberbullying and cybervictimization. Our findings indicate that both CU traits and narcissism are associated with the onset of cyberbullying and cyber-victimization. Future research should further explore the mechanisms that perpetuate such effects (e.g., reduced prosocial emotions and self-esteem) in order to design effective interventions that target adolescents with psychopathic traits. Given that the majority of prior work focuses on cyberbullying perpetration, this finding highlights the importance of also exploring the potential influence of psychopathic traits on adolescents’ vulnerability to online victimization. Importantly, we found that family support was the only contextual factor that protected adolescents from cyber-victimization across time. However, it remains to be seen why this effect was specific for girls and not boys. By applying the ecological model of development, the current study contributes to a better understanding of which individual and contextual variables influence cyberbullying and cyber-victimization, and findings can have major theoretical and practical implications.

Disclosure statement

No potential conflict of interest was reported by the author(s).