You are on page 1of 14

Journal of Youth and Adolescence

https://doi.org/10.1007/s10964-018-0895-5

EMPIRICAL RESEARCH

Trajectories of Social and Emotional Competencies according to


Cyberbullying Roles: A Longitudinal Multilevel Analysis
1
Vítor Alexandre Coelho ●
Marta Marchante1

Received: 16 April 2018 / Accepted: 28 June 2018


© Springer Science+Business Media, LLC, part of Springer Nature 2018

Abstract
Current cyberbullying literature lacks longitudinal studies clarifying its predictors and consequences. This 1-year
longitudinal study investigated how social and emotional competencies develop according to Portuguese middle school
students’ involvement in cyberbullying, and whether class size influences this relationship. There were 455 participants
(Mage = 12.58; SD = 0.94; 46% girls), and data collection through self-reports took place in three different moments during
12 months. The results showed that students involved in cyberbullying in any role displayed negative trajectories during 1
year in self-control and social awareness, while victims and bully-victims displayed a more pronounced decrease in self-
1234567890();,:
1234567890();,:

esteem and relationship skills during the same period. Additionally, girls displayed higher initial social awareness levels,
while larger classes were associated with higher levels of self-control and responsible decision making. These results
supported the importance of conducting longitudinal research and using a multilevel approach to address this topic.
Keywords Cyberbullying Social and emotional competencies Middle school Class-level variables
● ● ●

Introduction teachers. The authors concluded that this is due to, with the
huge progresses and dissemination of digital technology.
Presently, there is a consensus among researchers that fur- For instance, in Portugal, almost all youths (99.6%) own a
ther longitudinal studies are needed to fully understand the mobile phone and 90% of adolescents aged 12–18 have
relationship between predictors and/or consequences of Internet access at home (INE 2016).
cyberbullying (Coelho and Sousa 2017; Nocentini et al. Cyberbullying can be defined as the deliberate and
2013; Savage and Tokunaga 2017). Among relevant vari- repeated misuse of electronic media (e.g., e-mail; SMS,
ables, social and emotional competencies are often identi- MMS, social networking sites, chat rooms, etc.) to threaten
fied as either possible predictors of cyberbullying roles or or harm another person who cannot easily defend them-
likely to be impacted by cyberbullying (Coelho and Sousa selves (Savage and Tokunaga 2017). Therefore a cyberbully
2017; Nocentini et al. 2013). Cyberbullying is a growing would be someone who inflicts this harm on others, whereas
problem among school-aged children (Brown et al. 2017), a cybervictim is the recipient of such behaviors. Someone
given that the way children and adolescents communicate who takes on both roles would be named a cyberbully-
has been subjected to radical changes during the last decade. victim. Additionally, unlike traditional bullying, cyberbul-
The rapid development of modern communication tech- lying reaches a far wider audience at a rapid speed, and is
nologies has made the Internet and mobile phone critical to not bound by time nor by physical space (Kowalski et al.
their youth social life (Sticca et al. 2013). Palermiti et al. 2014; Smith 2015). Smith (2015) concluded that cyber-
(2017) have pointed out that in the last decade cyberbul- bullying differs from traditional bullying in a number of
lying has received increased attention from researchers and aspects: (a) it depends on at least some degree of techno-
logical expertise; (b) it is primarily indirect rather than face-
to face; so there is some ‘invisibility’ of those doing the
bullying; (c) the perpetrator does not usually see the vic-
* Vítor Alexandre Coelho tim’s reaction, at least in the short term; (d) the breadth of
vitorpcoelho@gmail.com
the potential audience is increased; and (e) it is difficult to
1
Académico de Torres Vedras, Travessa do Quebra-Costas, escape from.
Torres Vedras 9 2564-910, Portugal
Journal of Youth and Adolescence

There is a lack of longitudinal studies addressing middle school girls were more often cybervictims (Bonanno
cyberbullying (Nocentini et al. 2013; Savage and Tokunaga and Hymel 2013; Olenik-Shemesh et al. 2012) or cyber-
2017). Nocentini et al. (2013) called for additional long- victims and cyberbully-victims than boys (Kowalski et al.
itudinal studies about the relationship between cyberbully- 2014; Pettalia et al. 2013). In Portugal, in two studies with
ing and its predictors, while Savage and Tokunaga (2017) students of similar ages to the current study, girls reported
critiqued the general reliance on cross-sectional data in higher levels of victimization (Coelho and Sousa 2017;
cyberbullying research. There are clear benefits for obtain- Matos et al. 2017).
ing longitudinal data, because it would allow clarify if it is For developmental differences we also find discrepant
the experience of being involved in cyberbullying that results reported in the literature. Brown et al. (2017) con-
decreases students’ social and emotional competencies, or if cluded that there was not a clear understanding of devel-
it those students who have lower levels of social and opmental differences in experiences of cyber victimization.
emotional competencies that are more likely to cyberbully The author attributed this to the relative recency of cyber-
or be targeted as cybervictims, or even if it is creating a victimization research. However, Campbell et al. (2013)
reciprocal process. found that 9th grade, 14-olds displayed higher levels of
cyberbullying than 6th grade students, aged 11 and 12. Also,
Prevalence and Correlates of Cyberbullying in two studies implemented in Portugal (Coelho and Sousa
2017; Matos et al. 2017) older students in the studies (ages
Cyberbullying has grown to become a serious public health around 14) reported higher incidences of victimization than
concern among adolescents (Savage and Tokunaga 2017). younger students (ages around 12).
On average about 15% of adolescents may be victimized
online (Modecki et al. 2014), and 12% could be perpetrators Social and Emotional Competencies and
(Ybarra and Mitchell 2004). In Portugal, there are very few Cyberbullying
published articles about cyberbullying that would allow for
an estimation of its prevalence. Even so, the prevalence of Different social and emotional competencies have been
cyberbullying seems to be lower than in the US. Matos et al. found to protect or promote children’s involvement in
(2017) reported that 7.6% of students have been victimized, cyberbullying (Busch et al. 2015; Cook et al. 2010). Pre-
and 3.9% have bullied others at least once over the last year, sently, in the field of social and emotional competencies, the
whereas Coelho et al. (2016) reported that 5.2% of students social and emotional learning (SEL) framework has gar-
had been victimized while 2.8% of students admitted to nered a lot of attention, with over 500 articles published
being cyberbullies. over the last 20 years (Weissberg et al. 2015). The Colla-
Involvement in cyberbullying is associated with several borative for Academic, Social and Emotional Learning
negative health and psychological outcomes. Students who (CASEL 2013) organizes social and emotional compe-
were cyberbullied showed signs of elevated depression tencies into five interrelated key competencies: Self-
(Bonanno and Hymel 2013; Ybarra and Mitchell 2004), awareness, self-management, social awareness, relationship
anxiety (Fredstrom et al. 2011), loneliness (Olenik-Shemesh skills, and responsible decision making. There is evidence
et al. 2012), significantly higher drug and alcohol abuse that these key competencies may be crucial in under-
(Patchin and Hinduja 2012), and cyberbullying victimiza- standing bullying roles given that some authors (Busch
tion has also been linked to suicidal ideation (Bonanno and et al. 2015; Cook et al. 2010) have reported that children
Hymel 2013). Cyberbullies also reported higher levels of who bully, and those who are bullied, lack social problem-
drug and alcohol abuse (Patchin and Hinduja 2012), solving skills and positive self-concepts, while other studies
depression and anxiety (Campbell et al. 2013; Ybarra and have reported that bullied students have poorer social and
Mitchell 2004) and lack of or reduced empathy (Campbell emotional adjustment (Cook et al. 2010; Jenkins et al.
et al. 2013; Steffgen et al. 2011). 2016), and more difficulty making friends and poorer
relationships with classmates (Wachs 2012) than non-
Individual Characteristics cyberbullied students.
A key element in self-awareness is self-esteem. Defined
There is no consensus in the literature regarding gender by Rosenberg (1965) as “a favorable or unfavorable attitude
differences in cyberbullying and cybervictimization during toward the self” (p. 15). The relationship between cyber-
middle school. On one hand, several studies (Badaly et al. bullying and self-esteem has not been completely explored
2013; Brown et al. 2017) found no gender differences in in previous studies (Patchin and Hinduja 2010). Most stu-
cyberbullying or cybervictimization during that period. On dies concluded that victims of cyberbullying have low self-
the other hand, Li (2006) found that boys were more fre- esteem (Coelho and Sousa 2017; Fredstrom et al. 2011;
quently cyberbullies, while other authors concluded that Palermiti et al. 2017; Patchin and Hinduja 2010), even after
Journal of Youth and Adolescence

controlling for school-based victimization (Fredstrom et al. than cyberbullies and students not involved in bullying.
2011). However, for cyberbullies the results are less con- Additionally, Pabian and Vandebosch (2016) concluded
sensual, Patchin and Hinduja (2010) concluded that cyber- that exposure to cyberbullying as a bystander seems to lead
bullies had lower levels of self-esteem than do non-involved to lower levels of empathic responsiveness towards cyber-
students. Additionally, regarding the impact of cyberbully- victims over time.
ing on self-esteem, Sticca et al. (2013) concluded that the It is understandable why relationship skills are relevant
role of self-esteem as a longitudinal risk factor for cyber- to the issue of cyberbullying. According to Weissberg et al.
bullying has yet to be explored. For the current study, using (2015) relationships skills make it possible to initiate and
longitudinal data will allow to test whether differences in maintain positive interpersonal relationships, to respect
self-esteem related to cyberbullying are caused and/or an social norms, and to have good communication skills.
effect of such behaviors. Wachs (2012) concluded that feeling unpopular and being
Another key social and emotional competence is self- friendless were risk factors for cyber victimization, while
management (CASEL, 2013). Self-management is the other authors (Schoffstall and Cohen 2011) found that
ability to monitor and manage one’s own emotions and cyberbullying perpetration was related to having less mutual
behaviors in a way that facilitates motivation and the friends and lower rates of social acceptability, and popu-
achievement of personal goals (CASEL, 2013). Some larity. Other authors (Badaly et al. 2013) concluded that
authors (Busch et al. 2015; Jenkins et al. 2016; Vazsonyi more popular youths were concurrently more cyber victi-
et al. 2012) have found that a lack of emotion management mized and more often cyberbullying perpetrators than their
and self-control makes students more prone to being bullied peers.
by their peers (Busch et al. 2015) and Li et al. (2016) According to the Collaborative for Academic Social and
confirmed that a higher probability of students with low Emotional Learning (2013), responsible decision making
self-control participate in cyberbullying. Additionally, involves a reflexive consideration of different choices,
individuals low in self-control seek immediate pleasure specifically taking into account the wellbeing of self and
without much consideration of long-term consequences of others. Currently, we are not aware of any studies analyzing
their behaviors or actions and therefore self-control has a the relationship between responsible decision making and
greater effect in cyberbullying spaces compared to tradi- cyberbullying.
tional bullying (Vazsonyi et al. 2012).
One of the most relevant social and emotional compe- Class Characteristics
tencies in addressing cyberbullying is social awareness
(Coelho and Sousa 2017). Given that social awareness The most salient context in which bullying takes place in
consists of understanding other people, empathy, compas- childhood and adolescence is schools. Bullying is particu-
sion and norms, a lack of this competence can lead children larly frequent among classmates (Sentse et al. 2015), given
to act more hostile and aggressively in social situations, that the classroom is one of the most important social
such as by bullying others (Nickerson et al. 2008). contexts for psychosocial development (Söderberg et al.
Empathic concern has been considered as an important 2017). Classes are social settings of which students are
predictor of cyberbullying perpetration (Ang and Goh 2010; involuntary members, where they spend most of their time
Schultze-Krumbholz and Scheithauer 2013; Steffgen et al. at school and interact with other students daily (Sentse et al.
2011). Greater awareness of others’ feelings may not only 2015), it is therefore not surprising that some authors
allow students to treat each other with respect, but also reported that a considerable amount of the variance in
cause them to intervene and stop bullying situations bullying was due to differences between classes (Coelho
(Nickerson et al. 2008). However, cyberbullies can remain and Sousa 2017; Söderberg et al. 2017). Therefore, in
more anonymous and distant than traditional face-to-face addition to individual-level variables, several features of the
bullies, and therefore, will less likely observe the immediate classroom may influence bullying prevalence (Söderberg
consequences of their behavior (Pettalia et al. 2013). et al. 2017).
However, research findings regarding the relationship
between cyberbullying roles and empathy are inconsistent:
while Steffgen et al. (2011) found that cyberbullies showed Current Study
less empathy for cybervictims than those who have become
both cyberbullies and victims, and those not involved in The current study employs a multilevel longitudinal design
cyberbullying, Pettalia et al. (2013) concluded that cyber- to analyze the evolution of students’ social and emotional
victims and cyberbully-victims had higher levels of cogni- competencies (as conceptualized by Collaborative for
tive empathy than those not involved in bullying and that Academic Social and Emotional Learning 2013) during one
cyberbully-victims had higher levels of affective empathy school year, according to their role of involvement in
Journal of Youth and Adolescence

Table 1 Social and emotional


Characteristic Total (%) Self- Self Social Relationship Responsible
competencies across gender and
esteem control awareness skills decision making
grades
M (SD) M (SD) M (SD) M (SD) M (SD)

Gender
Male 246 37.02 12.43 10.28 (3.17) 8.59 (2.75) 6.23 (2.06)
(54.1%) (4.92) (2.78)
Female 209 37.44 12.93 11.79 (2.90) 8.09 (2.38) 6.29 (1.76)
(45.9%) (4.38) (2.53)
Grade
7th Grade 295 37.61 12.69 10.86 (3.28) 8.29 (2.69) 6.39 (2.05)
(64.8%) (4.45) (2.82)
8th Grade 160 36.47 12.61 11.20 (2.86) 8.48 (2.40) 6.01 (1.66)
(35.2%) (5.01) (2.40)
Note. N = 455

cyberbullying. Therefore, it tries to fill a gap in the present Methods


literature, given that previous research only analyzed the co-
occurrence of cyberbullying and cybervictimization with Participants
social and emotional competencies. And, to our knowledge,
none of them used a multilevel design or the current study’s The sample was a convenience sample composed of 455
theoretical framework. middle school students (who at T1 were frequenting 7th and
Therefore, our first set of hypothesis and research ques- 8th grade). These students were from 25 classes in five public
tions focuses on social and emotional competencies as middle schools in the district of Lisbon. Students were aged
antecedents of cyberbullying roles previous findings in the between 11 and 15 years (Mage = 12.58, SD = 0.94), how-
extant literature. Following the extant literature we hypo- ever 88.8% of the students were aged 12 and 13. The sample
thesized that, when compared to students who are not consisted of 246 boys (54.1%) and 209 girls (45.9%).
involved in cyberbullying; students involved in any cyber- Classes varied in size, ranging from 13 to 25 students
bullying role will have lower levels of self-esteem (fol- (Mclasssize = 18.73; SD = 3.07). Classes were extremely
lowing Patchin and Hinduja 2010) and self-control homogeneous in terms of ethnicity (99.3% of students were
(following Li et al. 2016; Vazsonyi et al. 2012); while Portuguese, the remainder Brazilian), gender distribution (in
cyberbullies will display lower levels of social awareness 92% of the classes between 33.3 and 66.7% of the students
(following Steffgen et al. 2011). Given the conflicting were boys) and socioeconomic status (ranging from 38.3 to
results found in the literature for relationship skills (Badaly 42.1% of students receiving free or reduced lunches).
et al. 2013; Wachs 2012), and the lack of studies regarding Regarding school location, 210 students attended urban
responsible decision making, we also posed two research schools and 245 students attended rural schools. Additional
questions. Do students involved in cyberbullying differ information about the participants is displayed in Table 1.
from non-involved students in relationship skills? and in Regarding attrition, beyond the six parents that did not
responsible decision making? consent to student participation, five students did not parti-
We also wanted to analyze two issues seldom addres- cipate in the second assessment due to having moved to a
sed in the literature. The first of them is how social and school outside the school grouping, while the same occurred
emotional competencies may be influenced by class size, to an additional 16 other students in the third assessment.
based on Coelho and Sousa (2017) we hypothesized that
there are differences in social and emotional competencies Procedure
between different sized classes. Finally, given that pre-
vious studies (because of their cross-sectional nature) did Data collection took place in three different moments; in
not provide insight into whether reduced social and time 1 (T1, October 2012) for the initial levels of social and
emotional competencies are a cause or a consequence of emotional competencies and demographical information; in
cyberbullying and cybervictimization, we hypothesized time 2 (March, 2013) for the mid-school year levels of
that being involved in cyberbullying (in any role) will social and emotional competencies; and time 3 (T3, October
lead to a decrease in every social and emotional compe- 2013) for the final levels of social and emotional compe-
tence analyzed. tencies and cyberbullying roles. Both T1 and T3 occurred
Journal of Youth and Adolescence

less than a month after the start of the respective school year Cyberbullying
(2012/13 and 2013/14). Evaluation took place after
obtaining consent from school boards and parents; the study The cyberbullying scale of the Bullying and Cyberbullying
followed the Portuguese Association of Psychologists Behaviors Questionnaire (QCBC; Coelho et al. 2016) was
(OPP) ethical standards and was approved by the Psychol- used. The QCBC (Coelho et al. 2016) is a 34-item self-
ogy for Positive Development Research Center. At all report measure, including 8 open-ended items (e.g., “Where
evaluations, the questionnaires were administered by one of did the bullying take place?”), intended to assess bullying
the four project educational psychologists as a part of a and cyberbullying behaviors, In the current study, we only
screening process for inclusion in a Social and Emotional employed eight items; 6 of them corresponded to the two
Learning program. Participants completed the ques- scales that assess two different participant role behaviors in
tionnaires in their regular classroom setting, in the presence cyberbullying; cyberbullies and cybervictims. Each of these
of their teachers. The role of the psychologists was to read scales contain three items, the former describing the action
questionnaire instructions aloud, explain the study proce- of being a cyberbully (e.g., “I sent inappropriate, rude, or
dure, reassure students of the confidential nature of the threatening messages by e-mail or by instant messaging
study, and to assist participants who required help. Students threatening”; α = .83 in the current sample), and the later
took roughly 30 min to fill out the questionnaires per the experience of being a victim of cyberbullying (e.g.,
classroom. If a student was not present during that period “They posted unwanted videos or photos of me on the
the psychologist returned the following week (n = 16). internet”; α = .73 in the current sample). Both subscales
assess the frequency of experiences by asking participants
Measures to report how often the behaviors described in each item
occurred during the previous school year. The other two
Social and emotional competencies items used provide students with a definition of bullying
(i.e., frequent, power differential, and negative intent), and
The Social and Emotional Competencies Evaluation then asked them to rate the frequency of bullying and vic-
Questionnaire (QACSE; Coelho et al. 2015; Coelho and timization on five-point scale. All items were scored on a
Sousa 2016) was used. This self-report instrument for five-point scale (1 = Never happened; 2 = Once or twice
adolescents (11 to 16 years) consists of 39 items. For the during the school year; 3 = 2 to 3 times a month; 4 = Once
purpose of this study we used four (the ones that directly a week; 5 = Several times a week). For the two scales
measured social and emotional competencies) of the ques- (cyberbullying and cybervictimization) the average of the
tionnaire’s six subscales: self-control (7 items such as three items is computed for each student. The QCBC has
“When I want to talk, I wait for my turn”; α = .73); social been previously used in several studies (Coelho et al. 2016;
awareness (7 items such as “I am kind to other when I see Coelho and Sousa 2017; Coelho and Romão 2018), and
they are having problems”; α = .87); relationship skills (7 exploratory and confirmatory factor analysis results have
items such as “I like to coordinate group activities”; α supported its two-factor structure in a sample of 1039
= .71); and responsible decision making (4 items such as “I middle school students (Coelho et al. 2016). In the current
ponder several alternatives before making a decision”; α sample these two-factors explained 59.6% of the variance.
= .87). The items are presented as statements to be rated on
a four-point scale (A = never; B = sometimes; C = fre- Data Analysis
quently and D = always).
We adopted the criteria used in previous studies with the
Self-esteem same instrument (Coelho et al. 2016; Coelho and Sousa
2017; Coelho and Romão 2018) to classify students as
We used the global self-esteem scale (Portuguese version; cyberbullies, cybervictims or cyberbully-victims. Students
Fontaine 1991) of the Self-Description Questionnaire-II were classified as cybervictims or cyberbullies if, in the
(Marsh et al. 1983). This scale allows the evaluation of third assessment, they rated the item describing victimiza-
global self-worth. The scale consists of 10 items (e.g., “In tion (or bullying) with a value of ≥ 2 and also rated at least
general I am happy to be the way I am”) rated on a five- one (of three) cybervictim (or cyberbully) behaviors as
point scale (1 = false; 2 = mostly false; 3 = nor true or having occurred at least once or twice during the previous
false; 4 = mostly true; 5 = true). Five of these items are school year. Students identified as cyberbully-victims were
presented as negative statements and are reverse scored. The those who filled these criteria for both roles.
scale’s internal consistency has been shown to be adequate The analyses conducted took into account that students
for both the original and Portuguese version, with Cron- from the same class have a much bigger probability of pro-
bach’s α of .88 and .82 respectively. viding responses with a high degree of correlation Heck et al.
Journal of Youth and Adolescence

(2013). Given the clustered nature of the data, with 445 stu-

Not Involved

29.90 (4.28) 38.71 (6.38)

13.67 (3.50)

11.92 (3.30)

8.90 (2.57)
6.40 (1.74)
dents nested within 25 school classes, we used multilevel

n = 375
M (SD)
models. To test our research hypotheses, three-level models
were used. First, an unconditional model (Model 0) with no
predictors was run to test for between-class variance. The

9.18 (3.13) 8.60 (1.65)

9.45 (3.50) 9.10 (1.52)

9.36 (3.14) 7.00 (1.76)


5.82 (1.60) 5.00 (1.25)
intercept was used as a random effect in the models and no
M (SD)
Victim/
Bullies
n = 20
time-varying covariates were included in the models. An
advantage of multilevel models is that they allow for the
progressive and/or simultaneous adjustment at distinct levels
(Heck et al. 2013). After several analyses, the best fit was
M (SD)
Bullies
n = 22

(4.23)
35.09

achieved with a linear measure of time (thereby creating


model 1) and auto-regressive as the covariance structure for
9.29 (1.16)

9.97 (2.55)

6.92 (1.70)
5.61 (1.44)
level 1.
Not Involved Victims

M (SD)
Time 3

In the current study, grade, gender and cyberbullying role


n = 38

(5.10)
31.50

were included in the model as Level 2 (individual) pre-


dictors (Model 2). For Model 3, class size (level-3 predictor
31.80 (4.08) 38.19 (5.82)

13.32 (3.16)

10.10 (2.13) 11.71 (3.66)

8.41 (3.14)
6.18 (2.02)

at level-3; grand mean centered) was introduced to deter-


n = 375
M (SD)

mine its association with the outcome variables where


intraclass correlations indicated that a significant portion of
the variance occurred between classes. That was not the
9.90 (1.79)

9.82 (3.63) 7.60 (2.22)


5.64 (1.86) 5.50 (1.65)

case for social awareness and relationship skills and


M (SD)
Victim/

responsible decision making, where the Intraclass Correla-


Bullies
n = 20

tion show that there was no need to include a 3rd level


Table 2 Descriptive statistics - social and emotional competencies per cyberbullying role, for each assessment

(class) in the models, since that there is not sufficient var-


iance explained at that level ( < .05), following the sug-
M (SD)
Bullies
n = 22

(5.73)

(4.06)

(4.15)
35.00

10.36

10.73

gestion by Heck et al. (2013). Finally, a Model 3b was


created adding the cross-level interaction between time (a
level 1 predictor) and cyberbullying role (a level 2 pre-
6.89 (2.12)
6.03 (2.06)

dictor). After several analyses, Scaled Identity was chosen


Not Involved Victims

M (SD)
Time 2

n = 38

(5.29)

(3.15)

(2.93)
33.92

10.71

10.68

as the covariance type for Level 2 and 3. All analysis and


models were estimated using IBM SPSS Statistics for
36.20 (4.44) 37.09 (4.68)

10.40 (1.27) 12.62 (2.61)

9.91 (4.28) 11.10 (2.77) 10.86 (3.09)

Windows, Version 20 (IBM Corp, Armonk, NY). Heck


8.34 (2.56)
6.20 (1.86)

et al. (2013) note that when using the SPSS mixed model,
n = 375
M (SD)

the reference group for a variable entered as a factor is the


last category.
8.55 (3.08) 8.40 (2.46)
6.27 (1.68) 5.60 (1.84)
M (SD)
Victim/
Bullies
n = 20

Results

Descriptive Statistics and Intraclass Correlations


Participants (N = 455)

M (SD)
Bullies
n = 22

(3.10)

(3.82)
34.27

10.73

As displayed in a Tables 2, 4.8% of students were involved


in cyberbullying as a bully, 8.4% of the students as a victim,
8.45 (2.89)
7.00 (2.49)

and 4.4% as cyberbully-victims. Considering gender, 7.3%


Victims

M (SD)
Time 1

n = 38

(4.45)

(2.85)

(3.03)

of boys were involved as bullies, 4.5% as victims and 2.9%


37.58

12.84

12.53

as bully-victims, while 1.9% of girls were involved as


bullies, 12.9% as victims, and 5.7% as bully-victims. Stu-
Responsible decision

dents’ participation in cyberbullying roles was similar


Relationship skills

across grades, for perpetration (5.2% for 7th grade; 4.2% for
Social awareness

8th grade), victimization (8.8% for 7th grade; 7.2% for 8th
Self-control
Self-esteem

grade), and cyberbully-victims (4.6% for 7th grade; 4.0% for


making

8th grade).
Journal of Youth and Adolescence

Table 3 Multilevel model


Parameters Model 0 Model 1 Model 2 Model 3b
analysis models for self-esteem
Null Level 1: time Level 2: Level 3: class and cross-
individual level interactions

Estimates of fixed effects


Intercept 37.37 (0.37) 36.98 (0.38) 48.34 (0.61)*** 45.90 (4.82)***
*** ***

Time 0.39 (0.13)** 0.38 (0.18)* 0.79 (0.16)***


Grade −1.27 (0.67) −1.25 (0.68)
Gender (girls = 1) −0.31 (0.44) −0.27 (0.44)
Victims −1.58 (0.79)* 0.01 (0.81)
Bullies −3.21 (1.42)* −2.98 (1.45)*
Bully-victims −2.47 (1.48) −0.83 (1.53)
Victims × Time −3.78 (0.40)***
Bullies × Time −0.48 (0.71)
Bully-victims × Time −3.77 (0.74)***
Class size 0.10 (0.08)
Estimates of covariance parameters
Repeated measures 14.52 (1.67) 13.26 (0.66) 13.86 (0.72)*** 11.57 (1.56)***
*** ***

Intercept individual 16.46 (1.65) 18.66 (1.72) 17.21 (1.56)*** 16.70 (1.52)***
*** ***

Intercept class 2.05 (0.84)* 1.87 (0.85)* 1.29 (0.64)* 1.08 (0.67)
ICC .062 .055 .040 .037
2
R (within) .052 .080
R2 (between) .310 .422
Deviance (−2loglikelihood) 7660.664 7652.350 7574.320 7473.209
Δ-2LL 8.314** 78.030*** 101.111
Number of estimated 5 6 11 15***
parameters
Note: *p < .05; **
p < .01; ***
p < .001

In the unconditional models, the Intraclass Correlation students not involved with bullying. Class size was not a
Coefficient (ICC), representing the proportion of variance significant predictor of self-esteem, however there were
between classes indicated statistically significant variation significant cross-level interactions, as victims and bully-
between classrooms for self-esteem (ICC = .055), self-control victims displayed a more pronounced decreased in self-
(ICC = .081), and responsible decision making (ICC = .064). esteem during the analyzed period (as seen in Model 3b,
There was not sufficient variance to be explained at the class Table 3). Additionally, after adjusting the model for the
level in social awareness (ICC = .046), and relationship skills cross-level interaction between time and bullying role, the
(ICC = .008), so only Level-1 predictors were introduced for initial differences in self-esteem for victims was no longer
these variables, following Heck et al. (2013) significant. Altogether, the individual- and class-level vari-
ables accounted for 8% of between-individual variance and
Self-Esteem 42.2% of between-class variance in self-esteem.

The results for self-esteem are displayed in Table 3. The Self-Control


addition of the within-individual predictor (time) was sig-
nificant (as displayed in Model 1, Table 2). When between As displayed in Table 4, adding the within-individual pre-
individual variables were added in Model 2 (Table 3), dictor (time) was significant (as seen in Model 1, Table 4).
between individual and between-classes variances Furthermore, the addition of individual level predictors
decreased by, respectively, 5.2 and 31%. Involvement with (Model 2, Table 4) led to a decrease in between-individual
bullying was the only significant predictor of self-esteem in and between-classes variance of, respectively, 10.1 and
Model 2, as having been a victim or a bully was associated 17.2%. Among between-individual variables, gender and
with significantly lower levels of self-esteem compared to cyberbullying roles were a significant predictor of self-
Journal of Youth and Adolescence

Table 4 Multilevel model


Parameters Model 0 Model 1 Model 2 Model 3b
analysis models for self-control
Null Level 1: time Level 2: Level 3: class and cross-
individual level interactions

Estimates of fixed effects


Intercept 12.74 (0.23) 12.58 (0.24)*** 12.91 (0.38)*** 9.33 (1.10)***
***

Time 0.21 (0.08)** 0.27 (0.07)*** 0.52 (0.07)***


Grade 0.26 (0.43) 0.12 (0.37)
Gender (girls = 1) −0.50 (0.24) *
−0.47 (0.25)
Victims −2.25 (0.44)*** 0.01 (0.50)
Bullies −3.00 (0.78)*** −1.65 (0.90)
Bully-victims −3.11 (0.82)** −1.61 (0.94)
Victims × Time −2.30 (0.24)***
Bullies × Time −1.30 (0.43)**
Bully-victims × Time −1.43 (0.46)**
Class size 0.19 (0.06)**
Estimates of covariance parameters
Repeated measures 4.78 (0.50)*** 4.69 (0.50)*** 4.45 (0.45)*** 3.67 (0.45)***
*** *** ***
Intercept individual 4.82 (0.38) 4.85 (0.41) 4.36 (0.40) 3.76 (0.53)***
* * *
Intercept class 0.87 (0.37) 0.87 (0.36) 0.72 (0.31) 0.42 (0.22)
ICC .083 .084 .076 .054
R2 (within) .101 .106
R2 (between) .172 .506
Deviance (−2loglikelihood) 6294.532 6281.347 6229.472 6127.706
Δ-2LL 13.185*** 51.875*** 101.641***
Number of estimated 5 6 11 15
parameters
Note. *p < .05; **
p < .01; ***
p < .001

control, with boys reporting lower levels of self-control than the individual level variables emerged as significant pre-
girls (p = .042), whereas students involved in any role in dictors of responsible decision making (as seen in Model 2).
cyberbullying displayed lower levels of self-control. In However, the addition of the individual level variables (in
model 3b (Table 4), the addition of between-classes vari- Model 2, Table 5) led to a decrease of 13% in between-
ables and cross-level interactions between time and bullying classes variances. The addition of class size and cross-level
role showed that both class size (p = 003) and all the cross- interaction between time and bullying role, only led to one
level interactions were significant predictors of self-control. significant results (as seen in Model 3b, Table 5), i.e., after
In our sample, students from larger classes displayed higher adjusting for the cross-level interaction with time, the results
levels of self-control, whereas students involved in bullying showed that students who were victims of cyberbullying had
in any role (victims, bullies and bully-victims) displayed a significantly higher levels of responsible decision making,
significantly more pronounced decrease in self-control, but they also had a significantly more pronounced decrease
when compared to students not involved in cyberbullying when compared students not involved in cyberbullying.
during the analyzed time. After adjusting for the cross-level Altogether, the class- and individual-level variables explain
interactions, cyberbullying roles did no longer significantly 0.5% of the between-individual variance and 39.1% of the
predict self-control initial levels. Altogether, the between- between-class variance in responsible decision making.
individual and between-class level variables explained,
respectively, 10.6% of the between-individual variance and Social Awareness
50.6% of the between-class variance in self-control.
The results for social awareness and relationship skills are
Responsible Decision Making displayed in Table 6. Time, a within-individual variable,
was a significant predict of social awareness (as seen in
Table 5 displays the results for Responsible Decision Model 1, Table 6). As model 2b shows, adding the indi-
Making. Neither the addition of time (in model 1) nor any of vidual variables and cross-level interaction between time
Journal of Youth and Adolescence

Table 5 Multilevel model


Parameters Model 0 Model 1 Model 2 Model 3b
analysis models for responsible
Null Level 1: Time Level 2: Level 3: Class and Cross-
decision making
Individual Level Interactions

Estimates of fixed effects


Intercept 6.20 (0.12)*** 6.18 (0.13)*** 6.07 (0.22)*** 4.39 (0.67)***
Time 0.01 (0.04) 0.01 (0.04) 0.10 (0.04)**
Grade 0.30 (0.25) 0.22 (0.22)
Gender (girls = 1) −0.08 (0.16) −0.06 (0.16)
Victims −0.12 (0.28) 0.68 (0.31)*
Bullies −0.42 (0.50) −0.06 (0.55)
Bully-victims −0.64 (0.52) −0.19 (0.57)
Victims × Time −0.79 (0.13)***
Bullies × Time −0.32 (0.23)
Bully-victims × Time −0.40 (1.51)
Class size 0.09 (0.03)*
Estimates of covariance parameters
Repeated measures 1.23 (0.10)*** 1.23 (0.10)*** 1.23 (0.10)*** 1.01 (0.08)***
Intercept individual 2.10 (0.20)*** 2.10 (0.20)*** 2.10 (0.20)*** 2.09 (0.19)***
* * *
Intercept class 0.23 (0.11) 0.23 (0.11) 0.20 (0.10) 0.14 (0.08)
ICC .064 .064 .057 .043
R2 (within) .000 .005
R2 (between) .130 .391
Deviance (−2loglikelihood) 4932.203 4932.050 4928.149 4881.286
Δ-2LL 0.153 3.901 46.863***
Number of estimated 5 6 11 15
parameters
* ** ***
Note. p < .05; p < .01; p < .001

and bullying role led to a decrease of 6.1% in within-class (p < .001) or as a bully-victim (p < .001) led to a very sig-
variance and 5.9% in between-class variance. Gender and nificant decrease in the level of relationship skills.
the interaction between time and bullying role significantly
predicted social awareness, with girls reporting higher
levels of this competence (β = −1.51, SE = 0.27; t = Discussion
−5.52, p < .001), while bullies (p = 0.40) and bully-victims
(p = 0.17) displayed a significantly more pronounced The current study sought to address several important gaps
decline in social awareness than not involved students in cyberbullying research. The main of these was the lack of
during the analyzed time. However, during the same period, longitudinal studies focusing on the relationship between
victims (p < .001), displayed a much more significant cyberbullying roles and social and emotional competencies
decline in social awareness than students not involved in (Nocentini et al. 2013; Savage and Tokunaga 2017). This
cyberbullying. lack of longitudinal work precluded determining if reduced
social and emotional competencies (self-control, social
Relationship Skills awareness, self-esteem, relationship skills and responsible
decision making; CASEL 2013) are antecedents or con-
The addition of time (as displayed in Model 1) had a sig- sequences of cyberbullying roles (Sticca et al. 2013). The
nificant effect on relationship skills. Furthermore, the current study compared initial levels of competencies across
addition to the model of individual level variables and of the cyberbullying roles to test if social and emotional compe-
interaction between time and bullying role led to a very tencies lead to cyberbullying. However, it also compared
modest decrease (0.4%) in between-individual variance. their trajectories across cyberbullying roles which indicated
Gender significantly predicted relationship skills, with boys if reduced competencies are outcomes of cyberbullying
displaying higher initial levels (β = 0.48, SE = 0.24; t = roles. Furthermore, given that previous studies did not
2.00, p = .046) Involvement with cyberbullying as a victim contemplate how class size might influence the relationship
Journal of Youth and Adolescence

Table 6 Multilevel model


Parameters Social awareness Relationship skills
analysis models for social
awareness and relationship skills Model 0 Model 1 Model 2b Model 0 Model 1 Model 2b
Null Level 1: Level 2: Null Level 1: Level 2:
time individual and time individual and
cross-level cross-level

Estimates of fixed effects


Intercept 11.25 10.83 11.69 (0.36)*** 8.46 8.29 (0.14) 7.98 (0.26)***
(0.20)*** (0.21)*** (0.013)*** ***

***
Time 0.42 (0.06) 0.55 (0.06) 0.17 (0.05) 0.28 (0.05)***
*** ***

Grade 0.04 (0.40) 0.04 (0.27)


Gender (girls = 1) −1.51 (0.27) ***
0.48 (0.24)*
Victims −0.48 (0.54) 0.02 (0.46)
Bullies −0.33 (0.95) 0.45 (0.82)
Bully-victims −0.42 (1.00) 0.05 (0.86)
Victims × Time −0.97 (0.21) ***
−1.04 (0.15)***
Bullies × Time −0.77 (0.38) *
0.13 (0.28)
Bully-victims × −0.95(0.39)* −0.98 (0.29)**
Time
Estimates of covariance parameters
Repeated 4.29 (0.60) 3.52 (0.40) 3.19 (0.33)*** 2.01 (0.19) 1.92 (0.17) 1.65 (0.12)***
*** *** *** ***
measures
***
Intercept 6.23 (0.82) 6.92 (0.71) 6.50 (0.63) 5.50 (0.48) 5.59 (0.47) 5.57 (0.44)***
*** *** *** ***
individual
Intercept class 0.50 (0.28) 0.51 (0.28) 0.48 (0.26) 0.06 (0.12) 0.06 (0.12) 0.06 (0.12)
ICC .045 .046 .047 .008 .008 .009
R2 (within) .061 .004
R2 (between) .059 .000
Deviance 6326.089 6277.824 6210.741 5740.389 5725.956 5660.536
(−2loglikelihood)
Δ-2LL 48.265*** 67.083*** 14.433*** 65.420***
Number of 5 6 14 5 6 14
estimated
parameters
Note. *p < .05; **
p < .01; ***
p < .001

between cyberbullying roles and social and emotional Patchin and Hinduja 2010). However, in the current study,
competencies, whereas the current study analyzed the after adjusting the model for the cross-level interaction
aforementioned relationship both at individual and class- between time and cyberbullying role, there were no longer
room level. significant differences in the initial levels of self-esteem.
The results did not provide support for most of our first Therefore, the results support the proposition that cyber-
set of hypotheses, which focused on social and emotional victimization decreases students’ self-esteem, rather than
competencies as antecedents of cyberbullying roles. In the the proposition that students with lower self-esteem are
current study, only cyberbullies (neither cybervictims, nor more likely to be targeted as cybervictims.
cyberbully-victims) displayed lower levels of self-esteem. For self-control, however, the hypothesis we posed was
The results are in line with several other studies (Coelho not supported. After adjusting the models for the evolution
and Sousa 2017; Patchin and Hinduja 2010), with similar of self-control during the analyzed time, the current study’
age students, where cyberbullies had significantly lower initial levels of self-control are not significantly different
levels of self-esteem. Most studies had also concluded that between those students involved and not involved in
cybervictims displayed lower self-esteem (Coelho and cyberbullying. This result contradicts previous research
Sousa 2017; Fredstrom et al. 2011; Palermiti et al. 2017; (Busch et al. 2015; Jenkins et al. 2016; Vazsonyi et al.
Journal of Youth and Adolescence

2012). In the current study, students’ self-control levels only displayed a pronounced decline. Therefore, it appears that
decreased after their involvement in cyberbullying, sup- being cyberbullied affects the ability to consider different
porting the notion that online anonymity promotes disin- choices, specifically taking into account the wellbeing of
hibited conduct (as suggested by Suler 2004), and self and others.
contributing to lower self-control. Therefore, in cyberbul- An important feature of the current study was its multi-
lying, unlike traditional bullying, the importance of self- level design. Thus, we also analyzed if class size had any
control seems to be particularly relevant for all students’ influence upon social and emotional competencies. The
involved, and not only for cyberbully-victims. This is novel results partially supported this hypothesis, as larger classes
because previous findings concluded that angry and emo- are associated with higher levels of self-control and
tionally charged responses which tended to reinforce victi- responsible decision making. The association between lar-
mization (Skrzypiec et al. 2011) were a typical ger classes and self-control is in line with Coelho and Sousa
characteristic of bully-victims. Students involved in cyber- (2017), and it supports the notion that, in larger classes,
bullying in any role (not only bullies and bully-victims) students are required to develop higher self-control to
displayed more significant decreases in self-control, con- manage their behaviors in a way that allows them to achieve
firming that self-control is greatly affected by cyberbully- their goals (as proposed by Coelho and Sousa 2017).
ing, given the features of the internet, as suggested by Finally, we analyzed if involvement in cyberbullying led
Vazsonyi et al. (2012). to reduced levels of social and emotional competencies.
Regarding social awareness, our hypothesis was not And, indeed, the results supported that hypothesis, as stu-
supported by the results because cyberbullies did not dis- dents who were involved in cyberbullying displayed a
play lower levels of social awareness than non-involved negative trajectory in every social and emotional compe-
students. This finding contradicts several authors (Pabian tency analyzed during one school year. When compared to
and Vandebosch 2016; Schultze-Krumbholz and Schei- non-involved students, social awareness and self-control
thauer 2013) who had suggested that a lack of social decreased significantly for students involved in cyberbul-
awareness could be an important factor in leading children lying in any role (bully, victim or bully-victim), while
to act more aggressively in social situations. Cyberbullies victims and bully-victims also reported significant decreases
(and cyberbully-victims), however, reported significantly in their levels of self-esteem and relationships skills, and
more pronounced decreases in levels of social awareness victims also displayed a significant drop in responsible
than non-involved students, even though it was cybervic- decision making. As a whole, the results confirmed that
tims that reported the most pronounced decreases in social cyberbullying affects social and emotional competencies,
awareness. The latter is consistent with Pettalia et al. (2013) which is line with previous longitudinal cyberbullying stu-
who had proposed that those students who try to understand dies (Badaly et al. 2013; Pabian and Vandebosch 2016).
the perspective and emotions of others may be more
affected by hurtful cyberbehaviors. Furthermore, consistent Limitations
with Coelho and Sousa (2017), the current study’ results
show that girls displayed higher social awareness levels Although the current study employs a longitudinal
throughout the analyzed time. approach, something that is recommended by several
We also analyzed two research questions focusing on the authors (Coelho and Romão 2018; Pabian and Vandebosch
relationship between cyberbullying and two other compe- 2016; Schultze-Krumbholz and Scheithauer 2013), there are
tencies (relationship skills and responsible decision mak- several limitations that must be acknowledged. The main
ing). The results showed that in time 1 there were only limitation is that results are based solely on student’ self-
differences in cyberbullying roles for responsible decision reports. Even though for cyberbullying student reports are
making, with victims displaying higher levels of responsible valid for understanding the child’s perspective and some
decision making than bullies, bully-victims or not involved. authors (e.g., Hymel and Swearer 2015) have concluded
Additionally, and in line with Coelho and Sousa (2017) that self-reports are economical, efficient and give youth a
boys reported higher levels of relationship skills than girls. voice in the assessment process, since they tap into the
However, from time 1 to time 3, there were different tra- perceptions of both bullies and victims, this methodology
jectories according to cyberbullying roles in both compe- presents several weaknesses: they can be vulnerable to self-
tencies. In relationship skills, cybervictims and cyberbully- presentation strategies or influenced by social desirability
victims reported a significant decrease, which supports and memory biases (Hymel and Swearer 2015). A relevant
Wachs (2012), who had concluded that some elements alternative to self-reports is the use of peer-reports; some
relevant to relationship skills (e.g., feeling unpopular and authors have found this methodology to be highly reliable
being friendless) were risk factors for cyber victimization. in this context (Badaly et al. 2013; Wegge et al. 2016).
Also, for responsible decision making, cybervictims Therefore, future studies should include peer-reports and
Journal of Youth and Adolescence

teacher reports, even though teachers may be less aware of emotional competencies. Therefore, future studies should
cyberbullying than are students themselves. clarify the relationship between class size and both self-
Another limitation for this study is that it is impossible to control and responsible decision, as different efforts and
assert that students had no previous cyberbullying experi- strategies may be needed when developing anti-
ence before Time 1. Although this is an important discus- cyberbullying programs for different sized classes. Addi-
sion point, there are three main reasons to support the tionally, for self-control, future studies should try to identify
choice for the current study´s cut-point. First, elementary other class-level variables characteristics that need to be
students have less access to technology; especially within in controlled for, given that there was still significant variance
schools (the majority of elementary schools did not have to be explained even after adding class size as a predictor.
computers for students´ use). Second, self-report measures, Also, longitudinal designs should be employed to analyze
as used in the current study, must be used with children with potential gender ratio effects (i.e., concentration of boys
sufficient written and comprehension skills. Third, in Por- itself), another class-level characteristic, given the social
tugal, 7th grade is the entry point into upper middle school dynamics of classes with a higher concentration of boys that
and, as such, classes are reformed and there is a total may contribute to cyberbullying.
turnover of teachers (no 6th teacher remains with the stu-
dents), and previous reports concluded that such school
transition often result in an increase of bullying (Coelho and Conclusion
Sousa 2017).
Also, the number of bullies and bully-victims cast in the The current study examined the relationship between
current sample was smaller than desirable, even though cyberbullying roles and social and emotional competencies
cyberbullying prevalence rates were in line with other longitudinally during one year, while controlling for indi-
Portuguese studies such as Matos et al. (2017). These vidual and classroom level variables. Results showed that
authors reported that 7.6% of their sample had been students involved in cyberbullying displayed more pro-
cybervictims and 3.9% had been cyberbullies. The current nounced decreases in self-esteem, self-control, social
study’s prevalence rate is higher than the ones for the awareness, relationship skills and responsible decision
QCBC validation study, where Coelho et al. (2016) reported making than non-involved student. This pattern was most
that 5.2% of students had been cyberbullied and 2.8% of noticeable in social awareness and self-control, where
students admitted having cyberbullied other students. involvement in any cyberbullying role (bully, victim or
However, given the somewhat low incidence of cyberbul- bully-victim) led to more pronounced decreases. The find-
lying among Portuguese students, larger samples are ings suggest that low levels of social and emotional com-
necessary to more accurately describe the social and emo- petencies are more of a consequence of cyberbullying
tional competencies of students who assume those roles in involvement, rather than a cause for getting involved in
cyberbullying. cyberbullying. The current study’s results also showed that
students from larger class sizes had higher levels of self-
Future Studies control and responsible decision making. This finding
highlights the importance of controlling for class size (and
The current study also raised several topics that should be potentially other classroom level variables) in the analysis
addressed in future research. Future studies should employ of social and emotional competencies. Together, these
longer intervals of time, which means following students findings reveal the importance of strengthening students’
into Secondary school, to understand if continuous invol- social and emotional competences to prevent further nega-
vement in cyberbullying continues to lead to a drop in their tive health and psychological consequences of
levels of social and emotional competencies. Or, alter- cyberbullying.
natively, how long it takes students to reach a bottom level,
and if students who are no longer involved cyberbullying Acknowledgements The authors would like to thank Ana Maria
Romão, Patrícia Brás and Vanda Sousa for collecting and organizing
their previous levels of social and emotional competencies.
the data. We would also like to thank the students who took part in this
Also important for future studies would be to start mea- study. The present study was conducted following the national pro-
suring social and emotional competencies and cyberbully- fessional code of ethics for psychologists (OPP), following national
ing roles since elementary school because, according to the legislation.
Collaborative for Academic, Social and Emotional Learning Authors’ Contributions VAC conceived the study and its design,
(2013), this is a crucial time for social and emotional drafted the manuscript, performed the statistical analysis and partici-
competencies development. pated in the interpretation of the data; MM drafted the manuscript and
There is also a need to develop studies that further ana- participated in the interpretation of the data. Both authors read and
approved the final manuscript.
lyze the influence of class size upon several social and
Journal of Youth and Adolescence

Data Sharing and Declaration The datasets generated and/or analyzed Psychology International, 39, 121–137. https://doi.org/10.1177/
during the current study are not publicly available but are available 0143034317749992.
from the corresponding author on reasonable request. Coelho, V., Sousa, V., & Marchante, M. (2015). Development and
validation of the social and emotional competencies evaluation
Funding Project Positive Attitude is funded by Municipality of Torres questionnaire. Journal of Developmental and Educational Psy-
Vedras. chology, 5(1), 139–147. https://doi.org/10.5539/jedp.v5n1p139.
Coelho, V., Sousa, V., Marchante, M., Brás, P., & Romão, A. M.
(2016). Bullying and cyberbullying in Portugal: Validation of a
Compliance with ethical standards questionnaire and analysis of prevalence. School Psychology
International, 37, 223–239. https://doi.org/10.1177/
Conflict of interest The authors declare that they have no conflict of
0143034315626609.
interest.
Collaborative for Academic, Social, and Emotional Learning. (2013).
2013 CASEL guide: Effective social and emotional learning
Ethical approval The current study was approved by the Psychology
programs (Preschool and elementary school edition). Chicago,
for Positive Development Research Center.
IL: Weissberg, R. P., Goren, P., Domitrovich, C., & Dusenbury,
Informed Consent All school directors agreed to the implementation L.
of the programs and schools used passive informed parental consent, Cook, C. R., Williams, K. R., Guerra, N. G., Kim, T. E., & Sadek, S.
because the questionnaires were applied as a screening for potential (2010). Predictors of bullying and victimization in childhood and
inclusion into a program that is considered part of the school offering. adolescence: A meta-analytic investigation. School Psychology
Quarterly, 25, 65–83. https://doi.org/10.1037/a0020149.
Fontaine, A. M. (1991). Desenvolvimento do conceito de si próprio e
realização escolar na adolescência [Development of self-concept
References and school achievement during adolescence]. Psychologica, 5,
13–31.
Ang, R. P., & Goh, D. H. (2010). Cyberbullying among adolescents: Fredstrom, B. K., Adams, R. E., & Gilman, R. (2011). Electronic and
The role of affective and cognitive empathy, and gender. Child school-based victimization: Unique contexts for adjustment dif-
Psychiatry and Human Development, 41, 387–397. https://doi. ficulties during adolescence. Journal of Youth and Adolescence,
org/10.1007/s10578-010-0176-3. 40, 405–415. https://doi.org/10.1007/s10964-010-9569-7.
Badaly, D., Kelly, B. M., Schwartz, D., & Dabney-Lieras, K. (2013). Hymel, S., & Swearer, S. M. (2015). Four decades of research on
Longitudinal associations of electronic aggression and victimi- school bullying: An introduction. American Psychologist, 70,
zation with social standing during adolescence. Journal of Youth 293–299. https://doi.org/10.1037/a0038928.
and Adolescence, 42, 891–904. https://doi.org/10.1007/s10964- INE. (2016). Portugal em Números 2015 [Portugal in numbers 2015].
012-9787-2. Lisboa: Instituto Nacional de Estatística.
Bonanno, R. A., & Hymel, S. (2013). Cyber bullying and internalizing Jenkins, L. N., Demaray, M. K., Fredrick, S. S., & Summers, K. H.
difficulties: Above and beyond the impact of traditional forms of (2016). Associations among middle school students’ bullying
bullying. Journal of Youth and Adolescence, 42, 685–697. https:// roles and social skills. Journal of School Violence, 15, 259–278.
doi.org/10.1007/s10964-013-9937-1. https://doi.org/10.1080/15388220.2014.986675.
Brown, C. F., Demaray, M. K., & Secord, S. M. (2017). Cyber vic- Kowalski, R. M., Giumetti, G. W., Schroeder, A. N., & Lattanner, M.
timization in middle school and relations to social emotional R. (2014). Bullying in the digital age: A critical review and meta-
outcomes. Computers in Human Behavior, 35, 12–21. https://doi. analysis of cyberbullying research among youth. Psychological
org/10.1016/j.chb.2014.02.014. Bulletin, 140, 1073–137. https://doi.org/10.1037/a0035618.
Busch, V., Laninga-Wijnen, L., vanYperen, T. A., Schrijvers, A. J. P., Li, Q. (2006). Cyberbullying in schools: a research of gender differ-
& De Leeuw, J. R. J. (2015). Bidirectional longitudinal associa- ences. School Psychology International, 27, 157–170. https://doi.
tions of perpetration and victimization of peer bullying with org/10.1177/0143034306064547.
psychosocial problems in adolescents: A cross-lagged panel Li, C. K., Holt, T. J., Bossler, A. M., & May, D. C. (2016). Examining
study. School Psychology International, 36, 532–549. https://doi. the mediating effects of social learning on the low self-control
org/10.1177/0143034315604018. cyberbullying relationship in a youth sample. Deviant Behavior,
Campbell, M. A., Slee, P. T., Spears, B., Butler, D., & Kift, S. (2013). 37(2), 1–13. https://doi.org/10.1080/01639625.2014.1004023.
Do cyberbullies suffer too? Cyberbullies’ perceptions of the harm Matos, A. P. M., Vieira, C. C., Amado, J., Pessoa, T., & Martins, M. J.
they cause to others and to their own mental health. School D. (2017). Cyberbullying in Portuguese schools: Prevalence and
Psychology International, 34, 613–629. https://doi.org/10.1177/ characteristics. Journal of School Violence, 17(1), 123–137.
0143034313479698. https://doi.org/10.1080/15388220.2016.1263796.
Coelho, V. A., & Romão, A. M. (2018). The relation between social Marsh, H. W., Relich, J. D., & Smith, I. D. (1983). Self-concept: The
anxiety, social withdrawal and (cyber)bullying roles: A multilevel construct validity of interpretations basedupon SDQ. Journal of
analysis. Computers in Human Behavior, 86, 218–226. https:// Personality and Social Psychology, 45, 173–187. https://doi.org/
doi.org/10.1016/j.chb.2018.04.048. 10.1037/0022-3514.49.5.1360.
Coelho, V. A., & Sousa, V. (2016). Desenvolvimento de um sistema Modecki, K. L., Minchin, J., Harbaugh, A. G., Guerra, N. G., &
de avaliação multi-informantes de competências socioemocionais Runions, K. C. (2014). Bullying prevalence across contexts: a
[Development of a multi-informant evaluation system for social meta-analysis measuring cyber and traditional bullying. Journal
and emotional competencies]. In A. M. Pinto & R. Raimundo of Adolescent Health, 55, 602–611. https://doi.org/10.1016/j.ja
(Eds.), Avaliação e promoção de competências socioemocionais dohealth.2014.06.007.
em Portugal (pp. 135–153). Lisboa: Coisas de Ler. Nickerson, A. B., Mele, D., & Princiotta, D. (2008). Attachment and
Coelho, V. A., & Sousa, V. (2017). Class-level risk factors for bul- empathy as predictors of roles as defenders or outsiders in bul-
lying and victimization in Portuguese middle schools. School lying interactions. Journal of School Psychology, 46, 687–703.
https://doi.org/10.1016/j.jsp.2008.06.002.
Journal of Youth and Adolescence

Nocentini, A., Menesini, E., & Salmivalli, C. (2013). Level and Sticca, F., Ruggieri, S., Alsaker, F., & Perren, S. (2013). Longitudinal
change of bullying behavior during high school: A multilevel risk factors for cyberbullying in adolescence. Journal of Com-
growth curve analysis. Journal of Adolescence, 36, 495–505. munity & Applied Social Psychology, 23, 52–67. https://doi.org/
https://doi.org/10.1016/j.adolescence.2013.02.004. 10.1002/casp.2136.
Olenik-Shemesh, D., Heiman, T., & Eden, S. (2012). Cyberbullying Steffgen, G., König, A., Pfetsch, J., & Melzer, A. (2011). Are
victimisation in adolescence: Relationships with loneliness and cyberbullies less empathic? Adolescents’ cyberbullying behavior
depressive mood. Emotional and Behavioural Difficulties, 17, and empathic responsiveness. Cyberpsychology, Behavior, and
361–374. https://doi.org/10.1080/13632752.2012.704227. Social Networking, 14, 643–648. https://doi.org/10.1089/cyber.
Olweus, D. (1993). Bullying at school. Oxford: Blackwell. 2010.0445.
Pabian, S., & Vandebosch, H. (2016). An investigation of short-term Suler, J. (2004). The online disinhibition effect. Cyberpsychology and
longitudinal associations between social anxiety and victimiza- Behavior Journal, 7, 321–326. https://doi.org/10.1089/
tion and perpetration of traditional bullying and cyberbullying. 1094931041291295.
Journal of Youth and Adolescence, 45, 328–339. https://doi.org/ Vazsonyi, A. T., Machackova, H., Sevcikova, A., Smahel, D., &
10.1007/s10964-015-0259-3. Cerna, A. (2012). Cyberbullying in context: Direct and indirect
Palermiti, A. L., Servidio, R., Bartolo, M. G., & Costabile, A. (2017). effects by low self-control across 25 European countries. Eur-
Cyberbullying and self-esteem: An Italian study. Computers in opean Journal of Developmental Psychology, 9, 210–227. https://
Human Behavior, 69, 136–141. https://doi.org/10.1016/j.chb. doi.org/10.1080/17405629.2011.644919.
2016.12.026. Wachs, D. (2012). Moral disengagement and emotional and social
Patchin, J. W., & Hinduja, S. (2010). Cyberbullying and self-esteem. difficulties in bullying and cyberbullying: Differences by parti-
Journal of School Health, 80, 614–621. https://doi.org/10.1111/j. cipant role. Emotional and Behavioral Difficulties, 17, 347–360.
1746-1561.2010.00548.x. https://doi.org/10.1080/13632752.2012.704318.
Patchin, J. W., Hinduja, S. (2012). Cyberbullying: An update and Wegge, D., Vandebosch, H., Eggermont, S., & Pabian, S. (2016).
synthesis of research. In: In J. W. Patchin, S. Hinduja (Eds.), Popularity through online harm: The longitudinal associations
Cyberbullying prevention and response. (pp. 13–35). New York, between cyberbullying and sociometric status in early adoles-
NY: Taylor & Francis. . cence. Journal of Early Adolescence, 36, 86–107. https://doi.org/
Pettalia, J. L., Levin, E., & Dickinson, J. (2013). Cyberbullying: eli- 10.1177/0272431614556351.
citing harm without consequence. Computers in Human Beha- Weissberg, R. P., Durlak, J. A., Domitrovich, C. E., Gullotta, T. P.
vior, 29, 2758–2765. https://doi.org/10.1016/j.chb.2013.07.020. (2015). Social and emotional learning: past, present, and future.
Rosenberg, M. (1965). Society and the adolescent self-image. Prin- In: In J. A. Durlak, C. E. Domitrovich, R. P. Weissberg, T. P.
ceton, NJ: Princeton University Press. Gullotta, (Eds.), Handbook of social and emotional learning:
Salmivalli, C. (2010). Bullying and the peer group: A review. Research and practice. (pp. 3–19). New York, NY: Guilford
Aggression and Violent Behavior, 15, 112–120. https://doi.org/ Press. .
10.1016/j.avb.2009.08.007. Ybarra, M. L., & Mitchell, K. J. (2004). Online aggressor/targets,
Savage, M. W., & Tokunaga, R. S. (2017). Moving toward a theory: aggressors, and targets: A comparison of associated youth char-
Testing an integrated model of cyberbullying perpetration, acteristics. Journal of Child Psychology and Psychiatry, 45,
aggression, social skills, and Internet self-efficacy. Computers in 1308–1316. https://doi.org/10.1016/j.adolescence.2004.03.007.
Human Behavior, 71, 353–361. https://doi.org/10.1016/j.chb.
2017.02.016.
Schoffstall, C. L., & Cohen, R. (2011). Cyber aggression: The relation
Vítor Alexandre Coelho is a certified Specialist in Educational
between online offenders and offline social competence. Social
Development, 20, 587–604. https://doi.org/10.1111/j.1467-9507. Psychology (as recognized by Portuguese Professional Psychology
2011.00609.x. Association, OPP), and possesses a PhD in Educational Psychology by
Schultze-Krumbholz, A., & Scheithauer, H. (2013). Is cyberbullying the Faculty of Psychology and Educational Sciences of the University
related to lack of empathy and social-emotional problems? of Coimbra. He is the coordinator of Positive Attitude project and a
International Journal of Developmental Science, 7, 161–166. member of the Psychology for Positive Development Research Center.
https://doi.org/10.3233/DEV-130124. His research interests are social emotional learning, bullying and
Sentse, M., Veenstra, R., Kiuru, N., & ssivalli, C. (2015). A long- cyberbullying, professional issues and middle school transition. He is
itudinal multilevel study of individual characteristics and class- currently the president-elect of the International School Psychology
room norms in explaining bullying behaviors. Journal of Association.
Abnormal Child Psychology, 43, 943–955. https://doi.org/10.
1007/s10802-014-9949-7.
Skrzypiec, G., Slee, P. T., Murray-Harvey, R., & Pereira, B. (2011).
School bullying by one or more ways: Does it matter and how do
students cope? School Psychology International, 32, 288–311. Marta Marchante is a nationally-certified Psychologist (as
https://doi.org/10.1177/0143034311402308. recognized by Portuguese Professional Psychology Association,
Smith, P. K. (2015). The nature of cyberbullying and what we can do OPP), and also has a master degree in Cognitive-Behavioral Therapy
about it. Journal of Research in Special Educational Needs, 15 by the Faculty of Psychology of the University of Lisbon. She
(3), 176–184. https://doi.org/10.1111/1471-3802.12114. coordinates the project Ser a Brincar, has been part of the Positive
Söderberg, P., Korhonen, J., & Björkqvist, K. (2017). Psychosocial Attitude project team since 2009 and is also a member of the
maladjustment at student and classroom level as indicators of peer Psychology for Positive Development Research Center. Her research
victimization. Violence and Victims, 32, 842–857. https://doi.org/ interests are social and emotional school adjustment, bullying and
10.1891/0886-6708.VV-D-15-00166. cyberbullying, professional issues.

You might also like