Childhood Trauma and Cyberbullying Perpetration: The Mediating Role of Callous-Unemotional Traits

ABSTRACT

Throughout the years, cyberbullying has been explored and connected to experiences of childhood trauma (CT) and high levels of callous-unemotional (CU) traits. This study aims to explore differences between cyberbullying perpetrators and non-perpetrators and explore the mediate role of CU traits between CT and cyberbullying perpetration (CBP). A cross-sectional study was conducted with 7139 participants (3728 girls; 52.2%), aged 10–22 years, of whom 276 (59.8% male) self-reported cyberbullying perpetration. Propensity score matching was used to select a sample of 276 adolescents’ non-perpetrators with similar characteristics, in order to compare cyberbullying perpetrators with non-perpetrators. Results showed that CBP was more predominant among boys and middle adolescents. Callousness trait partially mediates the relation between all the studied dimensions of CT and CBP. Uncaring traits partially mediated the relation between emotional, physical, and sexual abuse. CU traits (i.e., callousness and uncaring) mediated the relation between physical, and sexual abuse, and physical and emotional neglect. This study gives a better understanding of the underlying mechanisms of the relation between CT and CBP, which allows professionals to have a different evaluation when considering intervention programs.

KEYWORDS:

• Childhood trauma
• callous-unemotional traits
• cyberbullying perpetration
• adolescence

With the wide use of electronic information and communication technology as a device of socialization in current times, the cyberbullying became an emerging issue that has been notorious among the adolescent population (Kim & Faith, 2020). Communication through electronic devices triggered an aggressive behavior between adolescents where its origins need yet a better understanding (George & Odgers, 2015). Cyberbullying is considered an intentional aggressive behavior perpetrated by an individual or group of, repeatedly (Barlett, 2019; Thomas et al., 2015; Tokunaga, 2010) which can take various forms such as threats via text or through sharing intimate content, through the medium of electronic devices (Hinduja & Patchin, 2008). Its high incidence made the need for theoretical and practical response more emergent than ever. However, even though studies conducted on the subject have been having a great progress into search for factor associated to cyberbullying (e.g., Hinduja & Patchin, 2008; Olweus, 1993), there is still lack of knowledge regarding which factors may contribute to increase this behavior among youth.

Accordingly with the General Aggression Model (GAM; Anderson & Bushman, 2002), there are processes that can contribute to the perpetration of aggressive behavior. This behavior can be molded affectively, cognitively, and physiologically, through the combination of (1) personal factors (i.e., biological predispositions, cognitive biases, mood, and personality traits), and (2) situational factors (i.e., provocation, exposure to violence, childhood trauma), in a specific context, that triggers the aggressive act (Allison et al., 2020; Plante & Anderson, 2017). An episode of acting aggressively can start a cycle of aggression thus, provoking a reaction on the perpetrator which feeds into an appraisal process that determines the course of action. This action feeds back into the cycle, influencing the perpetrator’s beliefs, attitudes, and expectations for future episodes/context (Anderson & Bushman, 2002; Plante & Anderson, 2017).

As mentioned by Zhu et al. (2021), risk and protective factors have been broadly studied, but confirmation is still needed that these factors can interact with each other, causing greater or lesser effects on cyberbullying among young people. Clarification of these issues would be useful to allow further research to recognize cyberbullying more accurately. In previous studies, childhood trauma (CT) was found to be a risk factor for cyberbullying perpetration (CBP) among adolescents (Hébert et al., 2016; Emirtekin et al., 2020). CT is associated with experiencing or witnessing situations that can involve abuse and neglect before the age of 17 (Kircaburun et al., 2020). This relationship can be explained by the Social Learning Theory (Bandura, 1977) which refers those children are capable of mimicking their parent’s behaviors, learn, and use them as an answer to solve their problems (Bae, 2021; Fanti et al., 2012; Khoury-Kassabri et al., 2016). According to several authors (e.g., Emirtekin et al., 2020; Kircaburun et al., 2020), children who experience childhood emotional abuse have a high chance to involve themselves in cyberbullying as perpetrators. In Zhang et al. (2020) study, it was found that childhood psychological abuse was significantly positively related to CBP, in college students.

The role of callous unemotional traits

Callous-unemotional (CU) traits refers to personal characteristics that are considered potential markers that might be the precursors for the development of aggressive behavior that reflects a disregard for others, lack of empathy, and guilt, and superficial or shallow expressions of emotions (Fanti et al., 2009; Frick & White, 2008; Kimonis et al., 2008; Munoz et al., 2012; Wright et al., 2019). Previous studies suggest that individuals who experience traumatic experiences during childhood tend to suppress their emotions becoming difficult to express, understand, and manage them (Wang et al., 2018; Xie et al., 2020). This relationship could be explained by the Adaptive Calibration Model (ACM) which proposes that individual characteristics developed as strategies aimed at adapting to stressful environments (Del Giudice et al., 2011). Thus, to deal with potentially traumatic situations, children can develop CU traits in order to protect against child abuse (e.g., Fang et al., 2020; Squillaci & Benoit, 2021). Based on the ACM, adolescents with more experiences of child maltreatment are more likely to report CU characteristics. In turn, CU traits have been considered strong predictors of aggressive behavior, especially cyberbullying (Orue & Calvete, 2019; Wright et al., 2019). In fact, several studies indicate that CU traits are positively associated with the perpetration of cyberbullying among community sample of adolescents (e.g., Fang et al., 2020; Fanti et al., 2012; Orue & Calvete, 2019; Wright et al., 2019). Despite the main objective and instruments resorted of the studies is not exactly the same, these studies explore de association between CU traits and cyberbullying perpetration. Results of Fanti et al. (2012) study revealed that CU traits were longitudinally related to cyberbullying and media violence exposure was a risk factor leading to both cyberbullying and cyber-victimization. Similarly, CU traits were positively related to cyberbullying at high levels of moral disengagement (Orue & Calvete, 2019). Specifically, a study of Wright et al. (2019) found that callousness and uncaring would correlate with cyberbullying perpetration and that online disinhibition would moderate the associations between both callousness and uncaring dimensions and both types of cyberbullying (i.e., non-anonymous and anonymous). Study of Fang et al. (2020) further extends previous research by confirming the mediating role of CU traits and the moderating role of perceived social support. The focus on CU traits brings additional nuances in linking childhood maltreatment to adolescent cyberbullying perpetration. However, these authors admit that further replication and extension are needed to understand, how childhood maltreatment relates to adolescent cyberbullying perpetration.

The present study

As the perpetration of cyberbullying has become a public health issue, it is important to understand its underlying characteristics. When understanding the development of cyberbullying, it is important to consider both individual traits and contextual variables. Given empirical evidence of the associations between childhood exposure to trauma and aggressive behavior problems, as well as emerging evidence of the association of these variables with CU traits. In the current study, we attempted to apply the previous knowledge and taking together the aforementioned models, were aims to explore the differences between perpetrators and non-perpetrators of cyberbullying and explore the mediate role of CU traits in relationship between CT and CBP (see Figure 1). Several studies indicate a mediating effect of CU traits on the relationship between CT and aggressive behaviors (e.g., Carlson et al., 2015; McDonald et al., 2017) and only one specifically reveals this effect on cyberbullying behaviors (Fang et al., 2020). Concerning to these last authors, childhood maltreatment had a greater impact on CU traits, through total scale, predicting cyberbullying perpetration for adolescents recruited from community with low levels of perceived social support. In this sense, the present study brought together the following hypotheses, based on the literature review: (H1) CU traits and each dimension (i.e., callousness and uncaring) would mediate the relationship between traumatic experiences during childhood and CBP; (H2) it is expected that CT experiences are associated with high levels of CU traits; (H3) therefore, it is expected that high levels of CU traits predict high levels of CBP.

Figure 1. Mediating effect of callousness, uncaring, and CU traits on the relationship between CT and CBP

Methods

Participants

The present data were collected from a part of the Interpersonal Violence Prevention Program (PREVINT^TM) which is an original psychological intervention program designed to prevent the development and expression of aggression during adolescence (Barroso et al., 2018). Initially, the sample was composed by 9028 adolescents. When asking highly sensitive questions (e.g., violent behaviors perpetration in daily lives), misreporting is frequently common and is often correlated with the level of social desirability. Research shows reports of perpetrating violence were more strongly correlated with social desirability (Sugarman & Hotaling, 1997), due to the need to respond in a culturally appropriate manner. People who are influenced by social desirability tend to over-report culturally desired behaviors and/or under-report undesired behaviors. The inclusion of a measure of social desirability measure was therefore considered in this sample prior to the data analyses. All participants were screened for social desirability, ruling out adolescents who scored over M = 14.73, which corresponds to two standard-deviation above de mean, as suggested by Almiro et al. (2016). After this process 1889 participants (20.92%) were removed due to a high score in social desirability scale, getting a sample composed by 7139 participants with ages between 10 and 22 years (M_age = 14.73, SD = 1.86) with 3411 boys (47.8%) and 3728 girls (52.2%) attending 52 public middle-and high-schools, in rural and urban areas, from various districts of Portugal, both mainland and islands. From those participants, 276 adolescents (59.8% male and 40.2% female; M_age = 14.47, SD = 1.67) self-reported cyberbullying perpetration through the following question: “Have you ever used the internet/cellphone to send texts/images with the intent of hurting/threaten/blackmail/deceit/disqualify/control someone?.” Propensity score matching (PSM) was used to select a sample of non-perpetrators, in order to compare cyberbullying perpetrators with non-perpetrators. PSM is a particularly useful method for randomly approaching subjects from different groups, reducing or eliminating selection bias and balancing covariates (participants’ characteristics) between groups. The PSM methodology implemented, and the most common, was one-to-one, in which pairs of subjects from each group are formed so that the matched subjects have similar propensity score values (Austin, 2011). A sample of 276 participants non-perpetrators (59.8% male and 40.2% female; M_age = 14.84, SD = 2.14) with similar characteristics were randomly selected through answering “no” to the same question and who had valid data on all the variables of interest. All students in the sample received the intervention after the first assessment. It should be noted that according to the main aim of the present study only the first assessment before the intervention process was considered.

Measures

Sociodemographic questionnaire

A sociodemographic questionnaire was used to gather personal and sociodemographic data of the participants of the study relevant for sample characterization. This instrument was composed with questions regarding sociodemographic information, such as, age, sex, school year, school location, romantic relationship-related questions, and family-related questions.

Social desirability scale

The Social Desirability Scale (SDS-20; Almiro et al., 2016) is a self-reported measure, composed of 20 items, of dichotomous response (yes/no; e.g., “Have you ever cheated in a game?”; “Do you sometimes put off until tomorrow what you should do today?”). Each item is quoted with 1 point if the answer is in the sense of social desirability, 0 points if it is in the opposite direction. Prior to data analyses, all participants were screened for social desirability, ruling out adolescents who scored over M = 14.73, which corresponds to two standard-deviation above de mean, as suggested by Almiro et al. (2016). This instrument evaluates the tendency to transmit socially desirable responses rather than choosing responses that were a true reflection of their behaviors or feelings (Grimm, 2010). Cronbach’s alpha α for total scale was .80.

Cyberbullying perpetration

Experiences of cyberbullying were assessed through a question regarding its perpetration “Have you ever used the internet/cell phone to send texts/images with the intent to hurt/blackmail/threat/control/disqualify/incapacitate someone?.” Answers were rated as 0 = No, 1 = Yes.

The inventory of callous-unemotional traits

CU traits were measured with The Inventory of Callous-Unemotional Traits (ICU; Essau et al., 2006) adapted for the Portuguese population (Carvalho et al., 2018) is a 19-item self-report questionnaire which includes assessing two dimensions: callousness (composed by eight items) and uncaring (composed by 11 items). Responses vary through along a 4-point Likert scale ranging from 0 (“not at all true”) to 3 (“definitely true”; Carvalho et al., 2018). In this study, the total Cronbach’s α was .82, Cronbach’s α for Uncaring dimension .88, and Callousness dimension α = .80.

Childhood Trauma Questionnaire

Childhood trauma was assessed by the Childhood Trauma Questionnaire (CTQ) developed by Bernstein et al. (2003), adapted for the Portuguese population (Dias et al., 2013). The instrument is constituted by a 28-items which describe childhood experiences of exposure to maltreatment. Participants answer on a five-point Likert scale from 1 (“never true”) to 5 (“very often true”). Childhood trauma was measure using five subscales: Emotional abuse (e.g., “I thought that my parents wished I had never been born”); α = .83, Emotional neglect (e.g., “I felt loved”; α = .84), Sexual abuse (e.g., “I believe that I was sexually abused”; α = .92), Physical abuse (e.g., “I believe that I was physically abused”; α = .85), and Physical neglect (e.g., “I don’t have enough to eat”; α = .66). Regarding the total scale, Cronbach’s α was .85. Each subscale contains five items, and an additional three items are intended to measure any tendency to minimize or deny the abuse (e.g., “There was nothing I wanted to change in my family” – item 10; “I had a great childhood” – item 16; “I had the best family in the world” – item 22). Our study did not include the analysis of the denial index.

Procedure

The present investigation requested institutional authorization of National Commission for Data Protection (CNDP) and an University Ethics Committee. Afterward, data collection took place through computer-based questionnaires that participants would complete, in quiet classrooms. The anonymity of the study was emphasized, and informed consent was obtained from the legal guardians and participants before data collection. Participants were informed that their participation was voluntary, and they could terminate the participation anytime they want.

Analytic plan

In the present study, we adopted a cross-sectional design. To proceed with the statistical treatment of the data, the Statistical Package for the Social Science (SPSS) program, version 26 was used. No missing data were found in the final data. After that, to analyzed the characteristics of perpetrators and non-perpetrators of cyberbullying, sociodemographic characteristics were gathered. Secondly, instruments were categorized, and dimensions were built in each instrument suitable for the total sample. To explore the differences between perpetrators and non-perpetrators of cyberbullying, a series of independent t-tests were calculated and analyzed to identify existing significant differences between groups regarding CT, CU traits, and CBP. Then, effect size (Cohen’s d) of each dimension was calculated to measure effect size of the group means. Pearson’s and Spearman’s r was calculated to verify existing correlations between variables and to choose the best approach to interpret the results. Pearson’s r was calculated to test if the studied variables of CT and CU traits were correlated, where 0 indicates no correlation and +1/-1 indicates a perfect correlation. The magnitude of the effect of Pearson’s r was considered r = .10 as small effect, r = .30 as medium effect, and r = .50 as large effect (Field, 2018). Regarding the correlations between CT and CU traits to CBP, Spearman’s r was calculated, since CBP is considered a dichotomic variable, with categorized answers as “yes” or “no.” Pearson’s r and Spearman’s r were calculated to understand the relationship between variables and to help choose the best approach for analysis. From the obtained results of the correlations, it was possible to establish that the best option to use to interpret the results was through the analysis of the mediation effects, since there was an existent significant relationship between the studied variables. Therefore, to test the study hypothesis, a series of analysis were conducted.

The PROCESS macro for SPSS (Model 4) was applied to examine the mediation effect of CU traits on the relationship between CT and CBP (Hayes, 2018). Nonetheless, Model 4 includes four models to establish the relationship between variables, which were tested before hand through the correlation analysis, namely: Model 1 – the predictor (CT) significantly predicts the outcome variable (CBP); Model 2 – the predictor (CT) significantly predicts the mediator (CU traits); Model 3 – the mediator (CU traits) significantly predicts the outcome variable (CBP); and Model 4 – the predictor most predict less strongly in model 3 than in model 1. The bootstrap confidence intervals (CIs) determine whether the effects in Model 4 are significant, based on 5000 random samples (Hayes, 2018). An effect is regarded as significant if the CIs do not include zero. One advantage of the bootstrap method is that it does not require the normal distribution assumption and thus provides a more powerful test than traditional methods based on formulas with a normality assumption (Hayes, 2018).

Results

Prevalence and sociodemographic factors

A total of 276 participants (n = 165 boys, n = 111 girls) adolescents with ages between 10 and 20 (M_age = 14.47, SD = 1.67) reported engaging in cyberbullying behavior whilst 276 (n = 165 boys, n = 111 girls; M_age = 14.84, SD = 2.14) did not perpetrate cyberbullying. Most adolescents who reported engagement in cyberbullying were children of married (or the legal equivalent) couples (n = 196, 71.0%) and frequented high schools from various parts of the country. The number of cyberbullying perpetrators were higher in grades corresponding to seventh (n = 78, 28.3%), eighth grade (n = 61, 22.1%), and ninth grade (n = 72, 26.1%) than other grades. Participants who belonged in a family of middle and upper-class (n = 178, 64.5%) self-reported more engagement in cyberbullying.

Preliminary results

Variance of CT and CU traits on CBP

An independent-samples t-test was conducted to determine whether there were differences in the instruments used between adolescents who perpetrated cyberbullying and those who did not (see Table 1). For CTQ and ICU it was possible to verify that, there were significant differences in scores of each subscale, with higher scores for cyberbullying perpetrators than non-perpetrators, except for uncaring traits and CU traits, which were found higher in non-perpetrators.

Table 1. Descriptive statistics and group differences for CT and CU traits on CBP.

	Cyberbullying Perpetration
Variables	No (n = 276) M ± SD	Yes (n = 276) M ± SD	CI 95%	t	p	Cohen’s d
Sex
Male	n = 165	n = 165
Female	n = 111	n = 111
Age	14.84 ± 2.14	14.47 ± 1.67
Childhood Trauma
Emotional Abuse	7.35 ± 3.31	9.56 ± 4.76	[-2.89, -1.52]	-6.32	.000	0.53
Emotional Neglect	9.02 ± 4.36	11.16 ± 4.48	[-2.95, -1.29]	-5.04	.000	0.42
Sexual Abuse	5.65 ± 2.25	6.98 ±4.10	[-1.88, -0.77]	-4.72	.000	0.40
Physical Abuse	5.75 ± 2.05	7.65 ± 3.89	[-2.40, -1.36]	-5.43	.000	0.60
Physical Neglect	6.44 ± 2.44	7.92 ± 3.79	[-2.00, -0.94]	-7.13	.000	0.46
Callous-Unemotional Traits
Total	27.97 ± 8.94	22.23 ± 8.34	[4.29, 7.19]	-6.34	.000	0.66
Callousness	5.77 ± 4.99	8.13 ± 4.67	[-3.15, -1.54]	-5.72	.000	0.48
Uncaring	22.20 ± 7.43	14.11 ± 6.88	[6.89, 9.29]	13.27	.000	1.13

Note: N = 552.

Associations between CT, CU traits, and CBP

To analyze the associations between CT, CU traits, and CBP, a series of correlations analysis were executed between the study variables to verify the existence of significative correlations (see Table 2). Therefore, the means, SDs, Pearson, and Spearman correlations for the study variables are reported in Table 1. It was possible to find that experiences of childhood trauma were positively and significantly correlated to Callousness and CBP. Uncaring traits were negatively and significantly correlated to emotional abuse, physical abuse, and sexual abuse, and CBP. CU traits were positively and significantly correlated to sexual abuse, emotional neglect, and physical neglect.

Table 2. Correlations between variables, means, and standard deviations.

Variables	1	2	3	4	5	6	7	8	9
^aChildhood Trauma
1. Emotional Abuse	-
2. Physical Abuse	.66**	-
3. Sexual Abuse	.50**	.77**	-
4. Emotional Neglect	.33**	.28**	.23**	-
5. Physical Neglect	.43**	.62**	.57**	.62**	-
^aCU Traits
6.Uncaring Traits	-.16**	-.13**	-.10*	.01	.02	-
7.Callousness Traits	.26**	.34**	.36**	.18**	.34**	-.11**	-
8. Total	-.00	.07	.10*	.11**	.20**	.84**	.44**	-
9. ^bCyberbullying Perpetration	.25**	.32**	.15**	.15**	.13**	-.23**	.14**	-.15**	-
M	8.46	6.70	6.32	10.09	7.18	18.15	6.95	25.11	50%
SD	4.24	3.24	3.37	5.06	3.27	8.23	4.96	8.11	--

Note: N = 552

^aCategory using Pearson’s r

^bCategory using Spearman’s r

*p < .05.

**p < .01

Testing for mediation effect

The Model 4 of the PROCESS macro was used to test the mediation effect of the studied variables. From the obtained results, it was possible to find a significant indirect effect of emotional abuse on CBP through callousness trait (Indirect effect = .02, SE = .00, 95% CI [.01, .04]) and uncaring traits (Indirect effect = .04, SE = .01, 95% CI [.02, .07]). Physical abuse and CBP were partially mediated by callousness traits (Indirect effect = .03, SE = .01, 95% CI [.01, .06]) and uncaring traits (Indirect effect = .04, SE = .01, 95% CI [.02, .08]). The relationship between sexual abuse and CBP was also partially mediated by callousness traits (Indirect effect = .04, SE = .01, 95% CI = [.02, .07]), uncaring traits (Indirect effect = .03, SE = .01, 95% CI = [.01, .06] and CU traits (Indirect effect = −.02, SE = .00, 95% CI [−.04, −.00]). Results showed that the relationship between emotional neglect and CBP was partially mediated by callousness traits (Indirect effect = .01, SE = .00, 95% CI = [.00, .03]) and CU traits (Indirect effect = −.01, SE = .00, 95% CI [−.04, −.00]). Also, physical neglect and CBP were partially mediated by callousness traits (Indirect effect = .04, SE = .01, 95% CI = [.02, .07]) and CU traits (Indirect effect = −.05, SE = .01, 95% CI [−.09, −.03]). Hence, Hypothesis 1 was supported, where callousness trait partially mediated the relationship between all forms of CT and CBP; Hypothesis 2 was partially supported, where uncaring trait partially mediated the relationship between emotional abuse, physical abuse, and sexual abuse and CBP; Hypothesis 3 was partially supported where only sexual abuse, emotional neglect, and physical neglect were mediated by CU traits (see Figure 2).

Figure 2. Path analysis model of the relationship between childhood trauma and CBP mediated by callousness, uncaring, and CU traits

Solid lines indicate the existence of a significant relationship between variables; dotted lines indicate the nonexistence of a significant relationship between variables; Values are indicated above the respective relationship between variables.* p < .05** p < .001

Discussion

Cyberbullying perpetration (CBP) is considered an aggressive behavior perpetrated through online platforms, where the individual or group of individuals have the intent to hurt the victim, repeatedly (Tokunaga, 2010). Empirical evidence on cyberbullying has found that childhood trauma (CT) has an influence on the development of callous-unemotional (CU) traits and on cyberbullying perpetration (Emirtekin et al., 2020). Since most studies have very limited range of variables, it is pertinent to explore a new interaction between them, such as the study of mediation effects to obtain a better understanding of the reasoning behind CBP. Using the theoretical framework of the General Aggression Model (GAM; Anderson & Bushman, 2002), where individuals may develop their aggressive behavior due to the interaction between situational and personal factors, the current study explored the role of experiencing CT (i.e., emotional abuse, physical abuse, sexual abuse, emotional neglect, and physical neglect) and CBP mediated by CU traits (i.e., uncaring and callousness), in a sample of adolescents from Portugal. Therefore, the current study proposed that: Hypothesis 1 – callousness traits would mediate the relationship between CT and CBP; Hypothesis 2 – uncaring traits would mediate the relationship between CT and CBP; and Hypothesis 3 – CU traits would mediate the relationship between experiences of CT and CBP.

According with the obtained results, male adolescents (59.8%) self-reported higher levels of engagement in cyberbullying than female adolescents (40.2%), which is a common factor found among the literature, where male adolescents are related to a high expression of aggression and problematic behavior toward peers (Bae, 2021; Ciucci et al., 2014; Ciucci & Baroncelli, 2014; Hamal et al., 2019; Khoury-Kassabri et al., 2016; Li et al., 2019). Although, the use of technology has been growing among male and female adolescents, there is a tendency for CBP to turn equal and predominant for both sexes (Yudes et al., 2020). Also, CBP was more prevalent among the ages of 13 (23.2%) and 14 (25.0%), corresponding to the seventh (28.3%), eighth (22.1%), and ninth (26.1%) grade, which can be related to the moment that adolescents start to gain their own independence and, thus, the easier access to cyberworld, especially without parental and web restrictions, allowing adolescents to have freedom to misuse technology without repercussions (Govender & Young, 2018; Hamal et al., 2019; Yudes et al., 2020).

Through the analysis of the mediation effect on the relationship between CT and CBP it was found that callousness traits are partially accounted for that association, supporting Hypothesis 1. Empirical evidence supports these findings where CT can affect individual’s personality development (Bandura, 2001) and alter their way of perceiving themselves and the world around them in a negative and dysfunctional way (Goodman et al., 2017). According to some authors (e.g., Dackis et al., 2015; Fang et al., 2020; Kimonis et al., 2013; Squillaci & Benoit, 2021) children who experience CT tend to reveal the development of callousness traits as a coping mechanism to protect themselves from the environment they grow up in, characterized for being unpredictable and aggressive (Docherty et al., 2018). Taken these aspects into consideration, the General Aggression Model (GAM; Anderson & Bushman, 2002) gives a general understanding of this behavior, which indicates that cyberbullying can be influenced by personal (i.e., CU traits) and situational factors (CT). Individuals who experience CT and, consequently, develop callousness traits are more prone to engage in cyberbullying and, due to cyberbullying’s characteristics and the given appraisal of such actions, since cyberbullying can be perpetrated anonymously, enhances the probability to repeat the behavior. A possible explanation for this association is related to the fact that callousness traits are characterized by lack of empathy and guilt toward individuals which inhibits the perpetrator to understand the victim’s reaction to the behavior. Besides, cyberbullying’s characteristics, such as, anonymity, may enhance the probability for individuals to perpetrate this behavior, giving them opportunity to act without repercussions (Hinduja & Patchin, 2008). According to Bandura (1977), children who grow up in an abusive environment often develop hostile behaviors when socially interacting.

However, uncaring traits only mediated the relationship between emotional, physical, and sexual abuse and CBP, which partially supports Hypothesis 2. According to Bowlby (1973), children rely on their caregivers to interpret their behaviors and situations and learn how to behave in those moments when confronted with them. Children who are abused tend to create negative internal working models of the world and themselves, which can translate into an indifference regarding others and themselves. However, further studies are necessary to examine the influence of the uncaring traits by themselves on CBP amongst adolescent population.

Regarding CU traits, it was found that CU traits are considered a pathway that might explain the relationship between all forms of CT and CBP, except for emotional and physical abuse. Previous literature suggest that the presence of CU traits is considered a strong predictor of externalization of aggression, bullying, and consequently, CBP (Fang et al., 2020; Fanti et al., 2012; Orue & Calvete, 2019; Wright et al., 2019). Individuals who have high levels of CU traits often disregard others’ feelings and lack of care for themselves and others, which may influence the way they perceive their victims, disregarding their emotions not stopping them from getting to what they wanted initially, hurting the victim (Wright et al., 2019).

These findings are important for professionals when considering intervention programs: the use of psychoeducation in schools, involvement of parents and teachers that may appeal to adolescents’ empathic side, and group activities should be considered. Even though it might be difficult, sometimes, to detect abuse or neglect within family environment, teachers might play an important role in children, understanding their needs and minimizing the development of callous-unemotional traits. Also, creating a secure environment at school and at home for adolescents to be able to express themselves freely without repercussions through psychoeducation related to cyberbullying to parents to expand their knowledge and understanding of their children when presented with their concerns of their online activity.

Limitations and future directions

Several limitations were found in the current study. First, all variables were assessed through self-reported measures. Common to several empirical research, this study shared method variance bias that consists in a limitation as it potentially resulting in the artificial inflation of relationship of variables collected using the same respondents as a source for obtaining data (Jordan & Troth, 2019; Tehseen et al., 2017). Additionally, self-report approach could omit certain important data. Since Childhood Trauma Questionnaire involves questions regarding traumatic experiences during childhood, adolescents may have difficulty in recalling their memories. Nevertheless, this study resort to measures widely used in literature with community samples of adolescent (such as the Inventory of Callous-Unemotional Traits; e.g., Fang et al., 2020; Fanti et al., 2012; Wright et al., 2019). But it is noted that cyberbullying behavior is evaluated in a heterogeneous way with lack of consensus between the measures used across the studies. However, it appears that a substantial part of the studies that are based on cyberbullying, such as the present study, use a single item to identify the behavior. For future research, it should be taken into consideration the use of interviews where there is a possibility to a better understanding of the participants experiences and paying attention to behavioral cues. Third, due to the cross-sectional nature of the study, there is no possibility to infer causality, in this sense, future research could use a longitudinal method. Furthermore, studies using mediation effects might be crucial to understand underlying mechanisms and prevent cyberbullying behavior in adolescents.

Conclusion

In this study, it was possible to find that cyberbullying perpetrators are related to more experiences of CT and higher levels of callousness traits when compared to adolescents who do not have presence of high levels of aggressive behavior, in general. It is important to note that, adolescents who experience some type of trauma, related to abuse or neglect, can act more aggressively because of the influence of their CU traits. However, regarding uncaring traits, it was possible to establish that non-perpetrators had higher levels than perpetrators. For future studies, it is recommended to study the same mediation effect but in a wide sample of adolescents, to clearly understand this problematic and a better conceptual understanding regarding uncaring traits and its particularities, since evidence is still somewhat scarce.

Disclosure statement

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

Additional information

Funding

This work was supported by the Fundação Calouste Gulbenkian [PREVINT].