Translation paper

Misunderstood

journal_article.pdf

Home >Human Resource Management homework help >Translation paper

Human Communication Research ISSN 0360-3989

O R I G I N A L A R T I C L E

The Individual or the Group: A Multilevel Analysis of Cyberbullying in School Classes

Ruth Festl1, Michael Scharkow2, & Thorsten Quandt3

1 Department of Communication, University of Hohenheim, 70599 Stuttgart, Germany 2 Department of Communication, University of Hohenheim, 70599 Stuttgart, Germany 3 Department of Communication, University of Münster, 48143 Münster, Germany

In this study, we focus on the relevance of social influence to explain cyberbullying expe- riences among German high school students. Social influence is discussed in the context of computer-mediated communication. To obtain individual and sociostructural data, we conducted a survey study among German high school students (N = 4,282). Using multi- level modeling, we found that the attributes of the school class only contributed to the risk of being involved in cyberbullying to a small extent. Still, procyberbullying norms in class did enhance the risk of perpetration and victimization for students, even more so than their individual beliefs. Previous experiences with bullying and intensive, unrestricted use of the Internet were the strongest individual predictors of cyberbullying involvement.

Keywords: Cyberbullying, Computer-Mediated Communication, Social Norms, Peer Influence, Multilevel Analysis, Social Network Analysis.

doi:10.1111/hcre.12056

Recent reviews and meta-analyses of cyberbullying show that up to now, most studies have focused on psychological aspects and individual beliefs and attitudes (e.g., Slonje, Smith, & Frisen, 2013; Smith, 2012; Tokunaga, 2010). However, cyberbullying is by definition a social phenomenon: Following the description of traditional bullying by Olweus (1993) and Smith, Mahdavi, Carvalho, Fisher, and Russell (2008, p. 376), for example, define cyberbullying as “an aggressive, intentional act carried out by a group or individual, using electronic forms of contact, repeatedly and over time against a victim who cannot easily defend him or herself.” Moreover, like traditional bullying, cyberbullying frequently occurs between people who know each other outside the online context. Previous findings regarding cyberbullying show that the perpetrators and victims often know each other and come from the same school (Dehue, Bolman, & Völlink, 2008; Slonje & Smith, 2008). These parallels imply that attributes of the school class may not only be important in context of traditional bullying, but also in terms of explaining cyberbullying involvement.

Corresponding author: Ruth Festl; e-mail: [email protected]

Social Influence on Cyberbullying Involvement R. Festl et al.

Traditional bullying research already includes a large number of studies that pro- vide a sociostructural perspective on the phenomenon (e.g., Salmivalli, Huttunen, & Lagerspetz, 1997; Salmivalli, Lagerspetz, Björkqvist, Österman, & Kaukiainen, 1996). These approaches refer to different kinds of direct or indirect forms of social influ- ence within classrooms. Regarding the concept of indirect social influence, a certain amount of social resources are expected to enable or favor the perpetration of deviant behavior. This line of social influence was especially discussed in traditional aggres- sion research in terms of a person’s social position (see Neal, 2010). Direct social influence was not only considered in terms of injunctive classroom norms (classmates’ expectations about the acceptability of bullying, e.g., Salmivalli & Voeten, 2004), but also in terms of the actual behavior of relevant others (“descriptive norms,” e.g., Mout- tapa, Valente, Gallaher, Rohrbach, & Unger, 2004). However, this socio-structural line of research has only rarely been transferred to the context of cyberbullying.

When analyzing the role of classroom norms in the context of cyberbullying involvement, this local social influence appears in a more global, anonymous online environment. Previous studies have generally confirmed that, based on communica- tion and interaction, a strong social identity can also develop in computer-mediated groups in which visually anonymous individuals can easily exchange messages (e.g., Postmes, Spears, & Lea, 2000). Following the Social Identity Model of Deindividuation Effects (SIDE; Spears & Lea, 1994), social norms in these groups may exert a strong influence on the behavior of the members. Thus, an online environment without direct physical contact does not negate the forms of normative behavior; online communication and behavior can be even more strongly guided by social mecha- nisms than the individual features of the people involved. Moreover, Postmes et al. (2000) found that even preexisting groups, such as school classes, develop new ways of interacting when they enter a mediated context, which may dramatically change group dynamics and the social identity of its members.

Summarizing these considerations, we expected that individual beliefs and attributes, which have been the focus of most previous studies on cyberbullying, are even less salient in an online communicational environment, where cyberbullying behavior usually takes place. Following the general assumptions of SIDE theory, a sociostructural perspective may be a more promising approach when explaining cyberbullying experiences. In the present study, we therefore focus on the relevance of (injunctive and descriptive) classroom norms to explain cyberbullying experiences among German high school students. We analyze the role of sociostructural factors on cyberbullying involvement as compared to individual aspects, such as attitudes and personal experiences.

Previous research on cyberbullying

The individual level of influence Previous research on cyberbullying has mainly concentrated on the individual — for example, how differences in personality, attitude, experience, and Internet use relate to

R. Festl et al. Social Influence on Cyberbullying Involvement

cyberbullying behavior. Many of these individual factors can be confirmed as relevant levels of influence when explaining perpetration and victimization via the Internet (see Table 1).

Summarizing the previous findings, older adolescents seem especially likely to engage in cyberbullying, whereas there are no consistent results regarding gender dif- ferences as of yet. In line with the general findings on offline and computer-mediated behavior (see Reich, Subrahmanyam, & Espinoza, 2012), a strong overlap between traditional bullying and cyberbullying could be identified. This may be due to the basic underlying probullying attitudes that have previously been identified as relevant predictors of cyberperpetration (Salmivalli & Voeten, 2004). A strong influence of previous experiences was also identified by Walrave and Heirman (2009). They found that students who already perpetrated cyberbullying were also more likely to be vic- timized in the Internet. And, reversely, cybervictims had a higher risk of becoming a cyberbully. This might reflect general tendencies and motives of retaliation within the context of cyberbullying. However, these results were detected in a cross-sectional design, so that statements on causality could not be answered.

Only a few studies have analyzed the communication conditions and situations in which cyberbullying takes place. Slonje and Smith (2008) found that about one-third of victims did not know the gender and age of their bullies and that only 10% of the perpetrators were not from the same school. However, Dehue et al. (2008) also showed that although bully and victim typically attend the same school, most of the attacks are perpetrated from home, either alone or, to a lesser extent, with friends. Because many schools restrict the use of the Internet (see Smith et al., 2008), the perpetration of cyberbullying from home is not very surprising.

The relevance of the communication channel in the context of cyberbullying is also reflected by findings concerning adolescents’ Internet use. An obvious factor influencing cyberbullying is time spent online (e.g., Festl & Quandt, 2013; Walrave & Heirman, 2009). More intensive use of the Internet increases the risk of being involved in cyberbullying. However, some researchers contend that mere exposure is not con- vincing as a causal factor, especially for adolescents who are surrounded by digital media in their everyday lives (see Lenhart, Purcell, Smith, & Zickuhr, 2010).

In contrast, specific content-related aspects of media use are considered more important in explaining risk behavior online. Livingstone, Haddon, Görzig, and Olaf- sson (2011) showed that not only intensive use but also a varied array of online activ- ities affects the perpetration of risky online behavior. Other studies confirmed that the social features of the Internet, such as chat rooms and social network sites, are associated with a high risk of cyberbullying involvement (Walrave & Heirman, 2009; Ybarra & Mitchell, 2008).

Finally, the various opportunities for risky Internet behavior also depend on access to the actual equipment needed. Walrave and Heirman (2009) analyzed these media use conditions and found that perpetrators and victims slightly (but not significantly) more often than not had their own computers and could use them with little or no family supervision. Moreover, the distribution of mobile phones among

Social Influence on Cyberbullying Involvement R. Festl et al.

Ta bl

e 1

Li te

ra tu

re R

ev ie

w on

In di

vi du

al an

d So

ci os

tr uc

tu ra

lP re

di ct

or s

of Tr

ad iti

on al

B ul

ly in

g an

d C

yb er

bu lly

in g

Sa m

pl e

K ey

Fi nd

in g

So ur

In di

vi du

al pr

ed ic

to rs

A ge

N =

1, 50

1 yo

ut hs

;1 0–

17 ye

ar s;

48 % ♀

M or

e C

B am

on g

ol de

r st

ud en

ts Y

ba rr

a an

d M

itc he

ll (2

00 4)

N =

2, 05

2 st

ud en

ts ;1

0– 18

ye ar

s M

or e

C B

am on

g ol

de r

st ud

en ts

V an

de bo

sc h

an d

va n

C le

em pu

t( 20

09 )

M et

a- an

al ys

is C

ur vi

lin ea

r in

flu en

ce of

ag e

on C

B To

ku na

ga (2

01 0)

G en

de r

N =

1, 50

1 yo

ut hs

;1 0–

17 ye

ar s;

48 % ♀

N o

ge nd

er di

ffe re

nc es

in C

B Y

ba rr

a an

d M

itc he

ll (2

00 4)

N =

2, 05

2 st

ud en

ts ;1

0– 18

ye ar

s N

o ge

nd er

di ffe

re nc

es in

C B

Sl on

je an

d Sm

ith (2

00 8)

N =

1, 31

8 st

ud en

ts ;1

2– 18

ye ar

s; 49

% ♀

B oy

sm or

e oft

en cy

be rb

ul lie

s; gi

rl sm

or e

oft en

cy be

rv ic

tim s

W al

ra ve

an d

H ei

rm an

(2 00

9) N =

1, 21

1 st

ud en

ts ;Ø

13 ye

ar s;

50 % ♀

B oy

sm or

e oft

en cy

be rb

ul lie

s; gi

rl sm

or e

oft en

cy be

rv ic

tim s

D eh

ue et

al .(

20 08

)

A tt

itu de

s N =

1, 22

0 st

ud en

ts ;9

– 12

ye ar

s; 49

% ♀

Pr o-

bu lly

in g

at tit

ud es

po si

tiv el

y re

la te

d to

T B

bu lli

es Sa

lm iv

al li

an d

V oe

te n

(2 00

4) T

B N =

40 8

st ud

en ts

;1 2–

19 ye

ar s;

42 % ♀

T B

be ha

vi or

po si

tiv el

y pr

ed ic

te d

C B

be ha

vi or

, es

pe ci

al ly

fo r

bu lli

es Fe

st la

nd Q

ua nd

t( 20

13 )

C B

N =

1, 31

8 st

ud en

ts ;1

2– 18

ye ar

s; 49

% ♀

C yb

er bu

lli es

m or

e lik

el y

to be

vi ct

im iz

ed ;

cy be

rv ic

tim sm

or e

lik el

y to

cy be

rb ul

ly ot

he rs

W al

ra ve

an d

H ei

rm an

(2 00

9) In

te rn

et ac

ce ss

N =

1, 31

8 st

ud en

ts ;1

2– 18

ye ar

s; 49

% ♀

C yb

er bu

lli es

sli gh

tly m

or e

pr iv

at e

ac ce

ss to

th e

In te

rn et

(o w

n PC

) W

al ra

ve an

d H

ei rm

an (2

00 9)

So ci

al In

te rn

et us

e N =

1, 31

8 st

ud en

ts ;1

2– 18

ye ar

s; 49

% ♀

H ig

he r

on lin

e fr

eq ue

nc y

am on

g cy

be rb

ul lie

s W

al ra

ve an

d H

ei rm

an (2

00 9)

N =

1, 58

8 yo

ut hs

;1 0–

15 ye

ar s

So ci

al us

e of

th e

In te

rn et

as so

ci at

ed w

ith a

hi gh

er ri

sk of

be in

g cy

be rb

ul lie

d Y

ba rr

a an

d M

itc he

ll (2

00 8)

R. Festl et al. Social Influence on Cyberbullying Involvement

Ta bl

e 1

C on

ti nu

Sa m

pl e

K ey

Fi nd

in g

So ur

So ci

al po

si tio

n pr

ed ic

to rs

In de

gr ee

/c en

tr al

ity N =

40 8

st ud

en ts

;1 2–

19 ye

ar s;

42 % ♀

Po si

tiv e

co rr

el at

io n

be tw

ee n

ne tw

or k

ce nt

ra lit

y an

d be

in g

a C

B bu

lly /v

ic tim

Fe st

la nd

Q ua

nd t(

20 13

)

N =

1, 36

8 st

ud en

ts ;Ø

11 ye

ar s;

53 % ♀

In de

gr ee

ne ga

tiv el

y as

so ci

at ed

w ith

be in

g a

T B

vi ct

im M

ou tt

ap a

et al

.( 20

04 )

So ci

al pr

ef er

en ce

N =

46 1

st ud

en ts

;1 2–

13 ye

ar s;

49 % ♀

Pe rp

et ra

tio n

of T

B as

so ci

at ed

w ith

lo w

er so

ci al

pr ef

er en

ce C

ar av

ita ,D

iB la

si o,

an d

Sa lm

iv al

li (2

00 9)

Po pu

la ri

ty N =

57 3

st ud

en ts

;1 2–

13 ye

ar s;

50 % ♀

T B

vi ct

im sa

re ra

th er

pe rc

ei ve

d as

un po

pu la

r in

sc ho

ol Sa

lm iv

al li

et al

.( 19

96 )

N =

50 2

st ud

en ts

;1 0–

11 an

d 14

– 15

ye ar

s; 55

% ♀

T B

bu lli

es ar

e ra

th er

pe rc

ei ve

d as

po pu

la r

in sc

ho ol

Si jts

em a,

V ee

ns tr

a, Li

nd en

be rg

, an

d Sa

lm iv

al li

(2 00

9) Sc

ho ol

cl as

sp re

di ct

or s

T B

no rm

s in

cl as

s N =

1, 22

0 st

ud en

ts ;9

– 12

ye ar

s; 49

% ♀

A nt

i- bu

lly in

g no

rm si

n cl

as sn

eg at

iv el

y pr

ed ic

te d

pe rp

et ra

tio n

of T

B Sa

lm iv

al li

an d

V oe

te n

(2 00

T B

be ha

vi or

of cl

as sm

at es

N =

6, 98

0 st

ud en

ts ;5

0% ♀

R ej

ec te

d an

d so

ci al

ly an

xi ou

ss tu

de nt

sw ith

hi gh

er ri

sk of

T B

vi ct

im iz

at io

n in

cl as

sr oo

m s,

w he

re by

st an

de rs

re in

fo rc

e bu

lly in

K är

nä ,V

oe te

n, Po

sk ip

ar ta

,a nd

Sa lm

iv al

li (2

01 0)

T B

be ha

vi or

of fr

ie nd

s N =

1, 36

8 st

ud en

ts ;Ø

11 ye

ar s;

53 % ♀

Fr ie

nd s’

ag gr

es si

ve be

ha vi

or sp

os iti

ve ly

as so

ci at

ed w

ith be

in g

a T

B bu

lly M

ou tt

ap a

et al

.( 20

04 )

N =

45 9;

11 –

12 ye

ar s;

48 % ♀

Bu lli

es ar

e oft

en fr

ie nd

w ith

ot he

r bu

lli es

or su

pp or

te rs

,w he

re as

vi ct

im s’

ne tw

or ks

oft en

in cl

ud e

ot he

r vi

ct im

Sa lm

iv al

li et

al .(

19 97

)

C B

be ha

vi or

of fr

ie nd

s an

d cl

as sm

at es

N =

27 6;

13 –

19 ye

ar s;

43 % ♀

A hi

gh er

nu m

be r

of cy

be rb

ul lie

si n

sc ho

ol cl

as s

in cr

ea se

th e

ri sk

of vi

ct im

iz at

io n

Fe st

l, Sc

ha rk

ow ,a

nd Q

ua nd

t (2

01 3)

T B =

Tr ad

iti on

al bu

lly in

g; C

B =

C yb

er bu

lly in

g; Ø =

av er

ag e

va lu

e (e

.g .,

av er

ag e

in ye

ar s)

Social Influence on Cyberbullying Involvement R. Festl et al.

adolescents can also increase the risk of being involved in cyberbullying: These devices often guarantee unrestricted use because they are normally not shared among family members.

The sociostructural level of influence Looking at the sociostructural character of cyberbullying, there are two different dimensions to consider: indirect and direct influences. The former often relate to the social position of a person. This sociostructural concept is not immediately connected to cyberbullying, but it affects individual cyberbullying behavior by providing social resources.

Within this sociopositional approach, Neal (2010) differentiates between three concepts: social preference, network centrality, and perceived popularity. Measures of perceived popularity and social network position reflect prominence and social prestige within a peer group, whereas measures of social preference stress likeability (Neal, 2010, p. 124). These concepts are often measured using social cognitive map- ping techniques by creating a co-occurrence matrix based on peer nominations (e.g., Xie, Swift, Cairns, & Cairns, 2002).

The indicators therefore describe the social attributes of a person as perceived by their peers. Victims, in general, are known to be rather unpopular in school (Mout- tapa et al., 2004; Salmivalli et al., 1996), whereas the perpetration of bullying is often associated with power and dominance and therefore with the perception of a higher level of popularity (Sijtsema et al., 2009).

However, even though bullies are often perceived as popular, the perpetration of social aggression is also associated with a lower social likability (Caravita et al., 2009). Traditional victims were also found to have fewer friends in school (lower network centrality; see Mouttapa et al., 2004). However, all these results refer to traditional forms of bullying. Although one would expect that cyberbullying is experienced by adolescents who know each other, whether these social mechanisms work in an analo- gous way when analyzing bullying in the online context is not clear. In an initial study on social position and cyberbullying, Festl and Quandt (2013) showed that aggressive cybervictims who had already bullied someone else had more friends in school and occupied a central position in the network.

Although a person’s social status can be interpreted as an individual feature, it is provided through social relations with peers and therefore falls into the sociostruc- tural category. The availability or lack of social resources can be expected to make cyberbullying easier or more complicated, respectively, and thus encourages or pre- vents such behavior. In this study, we therefore control for a person’s social position when analyzing cyberbullying experiences.

Moreover, the social influence of peers can also be analyzed more directly by explicitly measuring and modeling forms of peer influence, such as peer pressure or behavioral learning (see Brown, Bakken, Ameringer, & Mahon, 2008). In general, there is a large body of research on adolescent behavior that focuses on peer influence, especially in the context of substance use (e.g., Sieving, Perry, & Williams, 2000) and

R. Festl et al. Social Influence on Cyberbullying Involvement

other forms of risky behavior (Jaccard, Blanton, & Dodge, 2005). Salmivalli et al. (1997) combined structural indices with individual bullying behavior and found that the social networks of perpetrators often consisted of other bullies or students who supported their aggressive behavior. The friendship networks of victims more often included other victims or persons that defended the victims. Mouttapa et al. (2004) also found that friends’ aggressive behavior was positively associated with being a traditional bully.

The term “peers” is generally used as a very broad catch-all concept that includes highly exclusive cliques, as well as peer crowds and other loose groups such as school classes (cf. Cotterell, 2007). Many studies have focused exclusively on close friends as the most important source of social influence and found evidence for this direct contact hypothesis (e.g., Haynie, 2001).

In contrast, other researchers argue that peer influence is not solely restricted to voluntarily chosen friendships. Adolescents are also directly or indirectly exposed to other peers (Payne & Cornwell, 2007). Juvonen and Galvan (2008) emphasized that the mechanisms of establishing social hierarchies and exerting social influence are particularly relevant in involuntary social groups, such as school classes. When stu- dents transition from elementary to higher schools, they are often confronted with a less structured system and a variety of new groups of classmates, triggering social processes such as deciding which classmates are “cool” and which are unpopular. Everyone must find his or her place in the social system of the class and is thus typically influenced by popular (and maybe bullying) classmates.

Additionally, challenging the behavior of bullies is thought to be rare, especially on the part of high-status students, because people are afraid of being bullied them- selves. This silent acceptance of bullying in class then reinforces probullying group norms that may not be representative of the class itself or of the individual class mem- bers (Juvonen & Galvan, 2008). This scenario illustrates the role of bystanders when analyzing the power of peers in the context of bullying. Passive behavior in bullying situations can be interpreted in different ways. Perpetrators can perceive the attention as additional support for their behavior, and victims assume that there is solidarity with the bully (Cowie, 2000; Salmivalli, Kaukiainen, & Voeten, 2005). Kärnä et al. (2010) confirmed that rejected and socially anxious students have a higher risk of vic- timization in classrooms where bystanders reinforce bullying instead of helping the victim (see Table 1).

Again, these findings come from studies that have focused on traditional forms of bullying. Following the so-called “bystander-effect,” even more passivity can be expected in the computer-mediated context of cyberbullying. This concept implies that the responsibility to help may be dispersed and therefore weakened in an online environment because of the larger number of bystanders involved in a particular situation. Markey (2000) confirmed these social mechanisms in a computer-mediated chat scenario. As the inclusion of a nearly unlimited audience is one of the most obvious characteristics of cyberbullying (e.g., Heirman & Walrave, 2008), the “bystander-effect” must be considered.

Social Influence on Cyberbullying Involvement R. Festl et al.

Summarizing these considerations, social influence in the context of cyberbullying may not be limited to close friends, but also include a more-or-less active audience that, especially in the case of social media, reflects the offline network of the bully. However, not only classroom norms on cyberbullying (see Salmivalli & Voeten, 2004) but also the classmates’ behavior may be relevant to individual cyberbullying risk. Festl et al. (2013), for example, found that having a higher number of cyberbullies in a school class increased the probability of individual victimization.

Research questions and hypotheses

In this study, we analyze whether the social influence of classmates is a relevant pre- dictor when explaining cyberbullying behavior, in addition to and perhaps even more greatly than individual beliefs and experiences. To evaluate the significance of the sociostructural level, we must first check for the relevant individual level of influence on cyberbullying involvement. As is known from sociopsychological research (e.g., Ajzen, 2005), a person’s behavior is strongly guided by his or her behavioral attitudes. Salmivalli and Voeten (2004) confirmed this relationship and found that probullying attitudes positively predicted the risk of becoming a traditional bully. We assume the same effect in the context of cyberbullying:

H1a: Procyberbullying attitudes are positively related to the risk of being a cyberbully.

Regarding previous bullying experiences, many studies confirmed a strong overlap between the perpetration and victimization of traditional bullying and cyberbullying behavior (e.g., Fanti, Demetriou, & Hawa, 2012; Festl & Quandt, 2013; Sticca, Ruggieri, Alsaker, & Perren, 2013). Moreover, in the context of cyberbullying, the findings also showed that perpetrating and experiencing cyberbullying were strongly related (e.g., Walrave & Heirman, 2009). Being a cyberbully therefore also enhances the risk of being victimized via the Internet and vice versa. Summariz- ing the previous findings on individual experiences, we therefore conclude the following:

H1b: Traditional bullying involvement is positively related to the respective cyberbullying involvement.

H1c: The perpetration of cyberbullying is positively related to the risk of being victimized in the Internet and vice versa.

Examining the sociostructural level, we generally expect that the mechanisms of social influence also operate in the context of computer-mediated communication. Previous findings on cyberbullying have confirmed that the involved students were typically from the same social environment, in many cases the same school. We therefore assume that offline social characteristics influence the online behavior of adolescents. However, following SIDE theory, group mechanisms and social influence should be even more pronounced in the online context because the visual anonymity and interchangeability of its members favor the group’s social identity (e.g., Postmes et al., 2000). This emphasis on the group in contrast to the singular individual is also

R. Festl et al. Social Influence on Cyberbullying Involvement

expected to be prevalent for preexisting social groups when interacting in an online environment.

Previous research on social influence in the classroom context has mainly focused on two concepts: the expected acceptability of a behavior within the classroom (injunctive classroom norm) and the actual behavior of the classmates (descriptive classroom norm; see Henry, 2001). Following this line of research, we also expect both forms of group norms to be relevant to the individual risk of becoming a cyber- bully and a cybervictim. Regarding victimization, previous findings have suggested that perpetrator and victim usually know each other from the offline world. Selecting a cybervictim, therefore, should follow the same social mechanisms that appear in traditional bullying.

According to the results of Kärnä et al. (2010), the risk of victimization was higher in classes in which bystanders supported the bully instead of the victim. In the con- text of computer-mediated communication, the probability of helping the victim is expected to be even lower (see Markey, 2000). Therefore, procyberbullying norms can be even more easily perceived as the predominant and desired way of acting. This can result in emotional or direct support for the perpetrators and therefore enhance the individual risk of victimization. Summarizing these considerations, we conclude the following:

H2a: Procyberbullying norms in class are positively related to the risk of being involved in cyberbullying as a perpetrator and a victim.

H2b: A higher number of traditional bullies and cyberbullies in class are positively related to the risk of being involved in cyberbullying as a perpetrator and a victim.

Finally, previous research on cyberbullying has almost exclusively concentrated on individual predictors. These individual-level predictors have been shown to be rele- vant factors influencing perpetration and victimization via the Internet. We therefore control for the most relevant predictors of cyberbullying, such as gender, age, educa- tion and Internet use. As additional social control variable, we also analyzed a person’s social position (see Table 1).

Method

Sample and study design To explain cyberbullying behavior on an individual and school class level, we con- ducted a comprehensive school survey in the southwest of Germany. Before recruiting the schools, we collected the consent of the Ministry of Education and the parents of the students. Altogether, 33 schools agreed to participate in our study. They covered the three different levels of education that are typical in Germany: lower-track educa- tion (“Hauptschule,” 10 schools), middle-track education (“Realschule,” 10 schools), and higher-track education (“Gymnasium,” 13 schools).

Because developmental processes occur quickly during adolescence, and we were striving to obtain a homogeneous sample, we only recruited students between the 7th

Social Influence on Cyberbullying Involvement R. Festl et al.

and 10th grades. Moreover, this age group was identified as being the most relevant in the context of cyberbullying behavior (Tokunaga, 2010). Within this age group, we tried to reach as many classes and students as possible. For the class-level analyses, we included all classes with at least five participating students and obtained a sam- ple consisting of 284 classes: 39 classes were from the lower track (14%), 97 classes were from the middle track (34%), and 149 classes were from higher-track schools (53%). On average, the schools provided data for nine classes (SD = 5.1; Min. = 1; Max. = 17). The classes had an average size of 15 participating students (SD = 5.0), with a maximum of 26. Overall, 5,656 students filled out the questionnaire during lessons in school.

For this study, 4,282 students (76%), 51% of whom were female and who had an average age of 13.9 (SD = 1.3) years, completed all relevant parts of the question- naire. Twenty-eight percent of the students were from the 7th (n = 1,204), 33% were from the 8th (n = 1,398), 26% were from the 9th (n = 1,109), and 13% were from the 10th (n = 571) grades. Most respondents were from higher-track (n = 2,551; 60%) and middle-track schools (n = 1,330; 31%), whereas only 9% (n = 401) were attending a lower-track school. On average, the students reported moderate social Internet use (M = 1.5; SD = 0.8; scale 0 – 4), but most of them had unrestricted access to the Inter- net via their own mobile phones or home computers (86%). Data were collected in February and March of 2013.

Measures Individual predictors Procyberbullying attitudes. To measure the participants’ attitudes toward cyberbul- lying, we used semantic differentials rated on a 5-point scale. The participants used four adjective pairs to indicate their evaluation of cyberbullying: “foolish — wise,” “negative — positive,” “bad — good,” and “not funny — funny” (α = .78; ω = .78). A high level of agreement with the scale reflects procyberbullying attitudes.

Cyberbullying behavior. Previous studies on traditional bullying and cyberbullying either used a definition-based or a behavior-based measure of the phenomenon (e.g., Sawyer, Bradshaw, & O’Brennan, 2008). The former may be problematic in terms of social desirability, whereas questions about concrete behavior tend to provide higher prevalence estimations (Sawyer et al., 2008). In this paper, we used a mixed approach: First, we provided a brief definition of bullying and cyberbullying, including behav- ioral examples. Afterwards, actual cyberbullying behavior was measured based on variants of such behavior (following the classification of Vandebosch & van Cleemput, 2009). We asked the students about 11 behaviors or experiences and used a reference timeframe of the last 12 months. They rated their answers on a frequency scale ranging from 0 (“never”) through 1 (“once”) and 2 (“occasionally”) to 3 (“often”). Following this approach, we could fulfill the definitional criteria of repetition as proposed by Smith et al. (2008). Six of these items referred to the perpetration of cyberbullying, whereas the other five items covered forms of victimization. The wording and the sta- tistical details of the items are displayed in the results section (Table 2). If someone

R. Festl et al. Social Influence on Cyberbullying Involvement

Table 2 Description of the Cyberbullying and Victimization Scale

Items Mean SD

Proportion (% agreed at least

“occasionally”)

Cyberbully someone How often during the last year …

have you sent an insulting message to someone? .55 .79 14 have you written something insulting about a

person on a public website? .15 .48 4

have you uploaded embarrassing pictures or videos of someone in the Internet?

.08 .35 2

have you written a message to someone using a fake identity in order to embarrass him?

.20 .53 4

have you spread rumors about someone in the Internet (e.g., using Facebook)?

.08 .34 2

have you forwarded a personal message of someone to others without his or her knowledge?

.27 .65 8

Being cyberbullied by someone How often during the last year …

have you been insulted while using the Internet? .47 .79 12 did someone intentionally post embarrassing

pictures or videos of you? .15 .47 3

did you receive a message of someone who used a fake identity to embarrass you?

.27 .65 7

did someone spread rumor about you in the Internet (e.g., using Facebook)?

.29 .67 7

did someone forward personal information of you to others?

.33 .68 8

Perpetration-Score .33 .77 22 Victimization-Score .37 .85 22

Note: N = 4,282; Items were rated from 0 (“never”), 1 (“once”), 2 (“occasionally”) to 3 (“often”); perpetration (Min. = 0, Max. = 6) and victimization score (Min. = 0, Max. = 5) were built on base of a sum score and dichotomized for the proportion indices.

answered at least one of the six perpetrator items with “occasionally” or “often,” the person was classified as a perpetrator. The same procedure was used for creating the victim category.

Traditional bullying behavior. The perpetration of traditional bullying in school was measured in a similar fashion as cyberbullying. Participants reported how often during the last year they had experienced each of the four behaviors described (e.g., “How often during the last year have you spread rumors about a schoolmate”) on a frequency scale ranging from 0 (“never”) through 1 (“once”) and 2 (“occasionally”) to 3 (“often”). If someone reported performing one of the perpetration items at least

Social Influence on Cyberbullying Involvement R. Festl et al.

occasionally, that person was classified as a traditional bully. To measure victimization in school, a dichotomous variable was created with the same rule, using the single item “How often during the last year have you been teased or insulted in school?”

Class-level predictors Pro-cyberbullying norms. We included the procyberbullying attitudes of the students as class-level predictors by averaging the individual scores from all students in a class. Following this approach, we obtained a measure of the average (individual) accept- ability of cyberbullying within each class.

Number of bullies and cyberbullies per class. As a second measure of social influ- ence, we also analyzed the descriptive classroom norm. The actual behavior of the classmates was thus measured by counting the number of bullies and cyberbullies in class using the individual perpetration scores.

Control variables Sociodemographics. In addition to the gender of the participants, we also controlled for the education level and the grade level of the school classes. Grade level was preferred over the individual age of the students because both indicators (age and grade level) were strongly correlated. The education level was measured using dummy variables for the middle-track and higher-track education schools with the lower-track students being the reference group in each case.

ICT use. The use of information and communication technologies was measured in two ways. As mentioned in the literature review, the use of mobile phones with Internet access is high among adolescents, and it usually guarantees unrestricted media use. Being online without parental control is also more probable if an adoles- cent has his or her own computer. We therefore asked the students if they had their own mobile devices or computers that enable them to access the Internet. Moreover, we calculated a formative measure for the online activities of the participants. On a 5-point frequency scale, ranging from 0 (“never”) through 1 (“seldom”) and 2 (“oc- casionally”) and 3 (“often”) to 4 (“very often”), they answered eight items regarding the social aspects of their Internet use (e.g., “How often during the last year have you used the Internet to visit a forum or an open chat,” “ … to play online games together with others,” “ … to search for new friends”) (M = 1.5; SD = 0.8; α = .72). The items were partly adapted from Livingstone et al. (2011).

Social position predictors. The following predictors are based on various measures from social network analysis. To reconstruct the school or class networks, we asked the students to fill out several name generators. These were used to compute the following indicators of social position in the class or school. The indegree indicates a person’s prestige, quantifying the rank an actor occupies within a set of actors (Wasserman & Faust, 1994). It was measured using a simple friendship generator: Students were asked to list their best friends in school (limited to 10 nominations). The indegree thus reflects the number of nominations received from all respondents in a school. Rather than focusing on these school-based social indicators, we focused on the class

R. Festl et al. Social Influence on Cyberbullying Involvement

level and asked the students to nominate the three classmates they liked the most. Again, the number of nominations received reflects the social preference of a per- son. Compared to the indegree, this indicator measures social preference, which is not necessarily associated with friendship. This procedure has been used previously in numerous bullying studies (e.g., Caravita et al., 2009). Students likewise mentioned three the classmates they thought were the most popular ones in class. A larger num- ber of nominations indicate a higher degree of perceived popularity.

Data analysis As noted above, this study is focused on the influence of social context on the individ- ual behaviors of students. Therefore, we modeled cyberbullying behavior as a depen- dent variable on the individual level. Because we obtained hierarchically nested data, in which one student can only be part of one specific class in one specific school, we applied multilevel analysis (Hox, 2010) with level 1 and level 2 predictors. The dependent variable in this study was the risk of becoming involved in cyberbullying behavior as a perpetrator or a victim. Because these were modeled as dichotomous variables, we applied the multilevel generalized linear model suggested by Hox (2010). As recommended by Enders and Tofighi (2007), we used group mean centering for the individual-level variables (level 1) and grand mean centering for the class-level variables (level 2). Missing data were removed listwise for each model. All analy- ses were conducted using the software R and the lme4 package (Bates, Maechler, & Bolker, 2012). When the analysis was completed, The regression coefficients were transformed into Odds Ratios (EXP(B)) for easier interpretation. Significance values are indicated as follows: **p < .01; *p < .05.

Results

Cyberbullying prevalence In the first step, we examined the prevalence of cyberbullying among the student sample. As explained above, we used a bullying and victimization scale to measure cyberbullying behavior. Except for one item, the scales were almost identical in cov- ering the active perpetration and the (passive) experience of cyberbullying. In general, agreement with the individual items was rather low, with sending insulting messages and forwarding personal information being the most common behaviors. Fourteen percent of the participants indicated that they had at least occasionally sent insulting messages to someone during the last year. Except for this item, all the other ques- tions received more support in terms of perceived victimization than perpetration. Altogether, 21.6% of the respondents had already cyberbullied someone in at least one of the described ways. Almost the same percentage was obtained for victimiza- tion (22.3%). Using both perpetration and victimization scores, we could identify 12% as perpetrators (without victimization experience), 12% as victims (without perpetra- tion experience), and a group of 10% of the respondents as having already experienced both (perpetrators/victims).

Social Influence on Cyberbullying Involvement R. Festl et al.

Of the 284 school classes included, we found only 2 to be completely unaffected by cyberbullying. Those classes that contained at least one cyberbully or one cybervictim included 4.2 bullies (maximum = 14) and 4.4 victims (maximum = 14) on average.

Individual-level predictors of cyberbullying involvement Perpetration risk First, briefly looking at the control variables, we can mainly confirm the previous find- ings on perpetrating cyberbullying (see Table 1). In line with earlier studies, we found a higher perpetration risk among older students (EXP(B) = 1.30**), and female, not male, adolescents were more strongly involved (EXP(B) = 1.85**). Additionally, stu- dents with intensive, unrestricted social use of the Internet showed an enhanced risk of perpetration. The respondents’ social position also affected their perpetration risk. A higher social preference by their classmates (slightly) positively predicted becoming a cyberbully (EXP(B) = 1.07**). Finally, we found that the individual perpetration of cyberbullying was less likely in classes of higher-track schools (EXP(B) = 0.58*). The findings are summarized in Table 3.

To answer our central research question, as a first step, we analyzed the relevance of individual beliefs and experiences in the context of perpetrating cyberbullying. In line with hypothesis H1a, procyberbullying attitudes increased the risk of individ- ual perpetration (EXP(B) = 2.01**). Therefore, in accordance with previous research explaining individual behavior, the perpetration of cyberbullying also seems to be strongly guided by individual beliefs. However, the strongest positive predictor was a previous experience as a traditional bully in school (EXP(B) = 4.63**; H1b con- firmed). Traditional bullies had an over four times greater risk of also perpetrating cyberbullying. This finding reflects a large overlap between both forms of bullying.

Finally, we found indications of retaliation behavior among students who had previously experienced victimization on the Internet (EXP(B) = 2.68**; H1c con- firmed). This mechanism could not be observed for traditional victims in school (EXP(B) = 1.02); only an online communicational context seems to increase the probability that victims also become perpetrators. The central findings are illustrated in Table 3.

Victimization risk As with perpetrating cyberbullying, the risk of victimization was also influenced by the included control variables. However, in contrast to previous studies, we did not find an effect for gender. Again, older (EXP(B) = 1.25**) and lower-track students (middle-track education: EXP(B) = 0.65*; higher-track education: EXP(B) = 0.53**) who intensively used Internet social platforms (EXP(B) = 2.12**) had an increased risk of being victimized on the Internet, whereas parental control of online use did not have a significant effect. Regarding the social position of a person, we also found a positive effect for social preference within a class (EXP(B) = 1.07**). Being liked by classmates, therefore, was not only associated with a higher risk of perpetration, but also with a higher probability of becoming a cybervictim.

R. Festl et al. Social Influence on Cyberbullying Involvement

Table 3 Explaining the Risk of Cyberbullying Involvement by Individual- and Class-Level Predictors

Perpetrator Victim

Fixed effects EXP(B) EXP(B) Control variables

Female 1.85**(a) 1.04 Grade level 1.30** 1.25** Middle track education 0.70 0.65* Higher track education 0.58* 0.53** Social Internet use 1.93** 2.12** Unrestricted Internet use 2.36** 1.23 Indegree in school 1.00 0.99 Social preference in class 1.07** 1.07** Popularity in class 0.95 0.99

Individual predictors Pro CB attitude 2.01** 1.13 TB perpetrator 4.63** 1.15 TB victim 1.02 4.54** CB perpetrator — 2.62** CB victim 2.68** —

Class level predictors Pro CB norm 5.36** 1.69* No of TB perpetrators 0.97 1.03 No of CB perpetrators — 1.01

Random effects σ2 Null model .31 .20 σ2 Final model .22 .14

Pseudo R2 .26 .18

Note: N = 4,282; Odds Ratios are indicated (EXP(B)); e.g., females have a 1.85 times higher risk of becoming a cyberperpetrator (value > 1 = enhanced risk), whereas students in middle track schools have a 30% lower risk (EXP(B) = 0.70; value < 1 = reduced risk) **p < .01; *p < .05; McFadden pseudo R2 was used (a) effect varies significantly across classes (random effect), likelihood ratio test; TB = Traditional bullying; CB = Cyberbullying.

Regarding cybervictimization, we expected the previous experiences of a person to be the most relevant predictors. In line with hypothesis H1b, we found that pre- vious experiences with traditional bullying as a victim were the strongest predic- tor of being victimized on the Internet (EXP(B) = 4.54**). Again, prior perpetration of traditional bullying did not coincide with a higher likelihood of cybervictimiza- tion, but previous cyberbullying increased the risk of becoming victimized online (EXP(B) = 2.62**; H1c confirmed). This finding again indicates mechanisms of retal- iation that may be prominent driving forces in the context of cyberbullying. However, due to cross-sectional nature of the data, we cannot test whether perpetration causes victimization or vice versa.

Social Influence on Cyberbullying Involvement R. Festl et al.

Class-level predictors of cyberbullying involvement Perpetration risk Following our central research question, we analyzed the role of norms and the behav- ior of classmates in an individual’s involvement in cyberbullying. In line with SIDE theory, a sociostructural perspective may be even more useful in a visually anony- mous online context (see Postmes et al., 2000). In general, the results of the multilevel regression analysis indicated that only a modest amount of variance in the risk of being involved in cyberbullying behavior as a perpetrator (9%) could be explained by the classes themselves (Intraclass Correlation Coefficient (ICC) = .09; see Hox, 2010, p. 128). Following these results, the class context seemed to play a rather subordinate role.

Although the general amount of variance on the class level is rather small, we found that classroom norms play an important role in predicting the individual per- petration risk. As postulated in hypothesis H2a, stronger procyberbullying attitudes in class increased the risk of individual perpetration (EXP(B) = 5.36**). Compared to the influence of individual attitudes (EXP(B) =2.01**), this social effect was even larger (overall effect of classroom norms: EXP(B) = 3.35**).

Because the average individual attitudes are negative and therefore reflect antibul- lying attitudes (M = 0.47; Min. = 0; Max. = 4), these effects can be interpreted as fol- lows: Even for students with average (anti)bullying attitudes in class, procyberbully- ing classroom norms enhance the risk of individual perpetration. This risk further increases if the person also has positive attitudes toward cyberbullying. Summarizing these findings, we conclude that offline classroom norms also play an important role in an online context, in addition to the relevant individual beliefs. Contrary to our expec- tations, the number of traditional bullies in class did not affect the individual behavior of cyberbullies (H2b rejected).1 Regarding the concept of descriptive norms, we found no significant effects. Therefore, previous individual experiences were clearly more important than the behavior of the relevant reference group.

Altogether, individual-level and class-level predictors, as well as control variables, explained more than half of the class-level variance and, in total, 26% of the variance in individual perpetration risk (summarized in Table 3). A series of likelihood-ratio tests indicated that only the individual effects of gender varied significantly across classes and that most such effects should therefore be modeled as random rather than fixed effects.

Victimization risk In the next step, we turned to those students who had been attacked via the Internet. The ICC for the null model (ICC = .05) showed that only 5% of the cybervictimiza- tion risk can be attributed to the class context. In order to explain this small class-level variance, we expected group norms and the behavior of classmates to be the rele- vant sources of influence. Again, we found a significant effect for procyberbullying classroom norms on the individual victimization risk (EXP(B) = 1.69*). Therefore, the probability of becoming a cybervictim is slightly higher in classes in which students

R. Festl et al. Social Influence on Cyberbullying Involvement

more positively evaluate cyberbullying. As expected, we identified no effects for atti- tudes on the individual level. However, as mentioned before, an individual’s previous bullying experiences play an important role in the risk of becoming a cybervictim. In contrast, the bullying behaviors of classmates did not influence this individual victim- ization risk. The social influence in an online communicational context, thus, seems to be restricted to the expected group norms, while the actual behavior of others rather seems to be insignificant.

Based on the specified model, we could explain 18% of the risk of becoming a cybervictim. The variance on the class level was rather low, and nearly half of it remains unexplained, despite the integrated class-level predictors in the model (see Table 3).

Discussion

In this article, we followed a sociostructural perspective in order to analyze students’ involvement in cyberbullying behavior. Specifically, we investigated how offline social influence on the class level affects the individual risk of being involved in online bul- lying as a perpetrator and a victim. Research on traditional bullying has already con- firmed the relevance of expected group norms (see Salmivalli & Voeten, 2004), as well as the actual behavior of classmates (see Salmivalli et al., 1997). However, up to now, this sociostructural perspective has been only rarely transferred to the context of cyberbullying. To account for the hierarchical data structure implied in this study, we applied a multilevel analysis with individual and class-level predictors. As a general result, we found that only a relatively small proportion of variance in cyberbullying could be explained by the class context. The risk of cyberbullying perpetration and victimization is largely determined by individual factors.

As a relevant predictor on the class level, we identified the cyberbullying-related classroom norms indicated by the average attitudes of the classmates. Being part of classes with high levels of procyberbullying norms increased the individual’s cyber- bullying risk, both as a perpetrator and a victim. This injunctive classroom norm, moreover, had a larger effect than the individual beliefs of a student. This result is especially important when analyzing the behavior of cyberbullies. If a student per- ceives procyberbullying norms as the predominant opinion in class, he or she does not necessarily have to support such a behavior strongly in order to perform it.

In contrast, we found no direct social influence for the classmates’ own bullying behavior. Neither the number of traditional bullies nor the number of cyberbullies in class affected the individual’s cyberbullying risk. However, due to the statistical endogeneity, we could not estimate the influence of the classmates’ cyberperpetration on the individual cyberperpetration in our cross-sectional design. Although the actual cyberbullying behavior of others does not seem to be influential, the individual involvement risk is strongly guided by the person’s own bullying experiences. Being a traditional perpetrator or a traditional victim in school strongly increased the respec- tive cyberbullying risks. Additionally, perpetrating or experiencing cyberbullying

Social Influence on Cyberbullying Involvement R. Festl et al.

were strong predictors of one another, while such cross effects could not be confirmed for the traditional bullying predictors. This seems to be in line with the assumption that status aspects lose relevance when communicating online because no longer are only socially inferior adolescents victimized (see, e.g., Dooley, Pyzalski, & Cross, 2009). However, a causal order within this relationship cannot be determined with the present cross-sectional data.

Summarizing these findings, we can indirectly confirm the “bystander effect” for cyberbullying as a computer-mediated form of communication. Not explicitly chal- lenging cyberbullies seems to result in a class climate in which at least indirect aggres- sion against others is tolerated or even supported. This in turn encourages students to actually perpetrate cyberbullying on their own and can result in a spiral of aggression and conformity in class. We can further conclude that norms negotiated in an offline group can influence an individual’s online behavior. This result suggests a strong over- lap between offline and online communication and the relevance of a sociostructural perspective on cyberbullying. Nevertheless, only the attitudes, not the actual behavior of the classmates, affected the individual cyberbullying risk.

The lack of behavioral influence for the classmates may be due to the fact that perpetrating cyberbullying can only seldom be observed directly. Social influence in an online context therefore seems to be guided by implicit perceived peer pressure rather than direct observational learning (see Brown et al., 2008). This result is in line with SIDE theory, which suggests the importance of social norms in mediated groups with a given social identity. Although school classes already exist in the offline world, it is expected that new ways of interacting are provided via communication online (see Postmes et al., 2000). Our results seem to suggest a strong orientation toward group norms, even stronger than that toward individual beliefs.

The social position predictors did not explain cyberbullying behavior very well. However, students who were more socially preferred by their classmates showed a higher risk of becoming involved in cyberbullying as perpetrators and victims. This finding contrasts with previous studies on social aggression and traditional bullying (Mouttapa et al., 2004; Salmivalli et al., 1996). The victims in our study did not fit into the classic role of an outsider normally found in traditional bullying research. Being socially accepted in class actually made the students more vulnerable to attacks via the Internet. This might be due to the composition of the victim group, which not only included “pure” victims but also persons who had already cyberbullied others. Festl and Quandt (2013) found a higher level of social prestige and centrality for those per- petrator/victims. Based on the current findings, we must also conclude that students with a higher social position — in the present case, a high level of peer acceptance in school — seem to be more involved in any kind of social interactions, whether these are positive (such as being invited to parties) or negative (such as being cyberbullied).

Despite these sociostructural findings, our results confirmed that cyberbullying involvement can still, to a large extent, be explained by individual-level predictors. As confirmed in previous studies, the risk of perpetration and victimization was higher among older and lower-track students who intensively use the Internet without

R. Festl et al. Social Influence on Cyberbullying Involvement

parental control. Surprisingly, we found that female students had a higher risk of being involved as perpetrators. This finding clearly contrasts with previous results that mainly confirmed male adolescents were more likely to be perpetrators and female adolescents were more likely to be cybervictims (Dehue et al., 2008; Smith et al., 2008). However, the effect is in line with traditional aggression research that found more frequent use of indirect and social aggression among girls (e.g., Björkqvist, Lagerspetz, & Kaukiainen, 1992). Thus, at least during adolescence, girls do not want to expose themselves explicitly as perpetrators who intentionally hurt other people. This may be due to gender-specific aspects of socialization. The technical features inherent in cyberbullying also enable the anonymous use of aggression, which seems to be the preference of girls.

Regarding general implications, we found that the school class (although the direct effects are rather small) is still a relevant concept for a better understanding of cyber- bullying and that class-level factors should also be considered in terms of preven- tion and intervention strategies. Classmates perceiving cyberbullying as being socially accepted in class seem to lower the barriers for the perpetration of cyberbullying. Compared to individual attitudes, this social effect was very strong. Thus, concentrat- ing on an individual person in order to prevent the appearance of cyberbullying is not sufficient. Even if a student does not endorse the perpetration of cyberbullying, being part of a class whose members socially accept the behavior increases his or her own risk of perpetrating and also experiencing cyberbullying. Successful strategies against cyberbullying, thus, must include a larger social group, in context of schools the whole school class. Normally, schools fine it impossible to change the general class composition in order to meet other risk factors such as higher diffusion among older and lower educated students. However, “soft” criteria such as classroom norms and attitudes can be more easily revised and modified.

In general, the relevance of social influence in context of cyberbullying was only confirmed for perceived group norms, not for descriptive norms, i.e., the classmates’ behavior. As already mentioned, this might be a specific characteristic of the computer mediated communication context of cyberbullying (see SIDE-theory). Therefore, we found it was not enough to solely transfer the considerations of traditional bully- ing research — highlighting the relevance of peers’ behavior — when integrating these social approaches. Future research needs to focus on the specifics of the communica- tion context when analyzing cyberbullying behavior.

Note

1 Due to the statistical problem of endogeneity, we could not check for the influence of the number of cyberbullies on the individual perpetration risk in a cross-sectional design.

References Ajzen, I. (2005). Attitudes, personality and behavior (Vol. ,2nd ed.). Berkshire, England: Open

University Press.

Social Influence on Cyberbullying Involvement R. Festl et al.

Bates, D., Maechler, M., & Bolker, B. (2012). lme4: Linear mixed-effects models using S4 classes (R package version 0.999999-0). Retrieved September 16, 2013, from http:// CRAN.R-project.org/package=lme4

Björkqvist, K., Lagerspetz, K. M. J., & Kaukiainen, A. (1992). Do girls manipulate and boys fight? Developmental trends in regard to direct and indirect aggression. Aggressive Behaviour, 18(2), 117 – 127. doi:10.1002/1098-2337(1992)18:2<117::AID-AB24801802 05>3.0.CO;2-3

Brown, B. B., Bakken, J. P., Ameringer, S. W., & Mahon, S. D. (2008). A comprehensive conceptualization of the peer influence process in adolescence. In M. J. Prinstein & K. A. Dodge (Eds.), Understanding peer influence in children and adolescents (pp. 17 – 44). New York: Guilford Press.

Caravita, S., Di Blasio, P., & Salmivalli, C. (2009). Unique and interactive effects of empathy and social status on involvement in bullying. Social Development, 18(1), 140 – 163. doi:10.1111/j.1467-9507.2008.00465.x

Cotterell, J. (2007). Social networks in youth & adolescence. London: Routledge. Cowie, H. (2000). Bystanding or standing by: Gender issues in coping with bullying in

English schools. Aggressive Behavior, 26(1), 85 – 97. Dehue, F., Bolman, C., & Völlink, T. (2008). Cyberbullying: Youngsters’ experiences and

parental perception. CyberPsychology & Behavior, 11(2), 217 – 223. doi:10.1089/ cpb.2007.0008

Dooley, J. J., Pyzalski, J., & Cross, D. (2009). Cyberbullying versus face-to-face bullying: A theoretical and conceptual review. Journal of Psychology, 217(4), 182 – 188. doi:10.1027/ 0044-3409.217.4.182

Enders, C. K., & Tofighi, D. (2007). Centering predictor variables in cross-sectional multilevel models: A new look at an old issue. Psychological Methods, 12(2), 121 – 138. doi:10.1037/ 1082-989X.12.2.121

Fanti, K. A., Demetriou, A. G., & Hawa, V. V. (2012). A longitudinal study of cyberbullying: Examining risk and protective factors. European Journal of Developmental Psychology, 9(2), 168 – 181. doi:10.1080/17405629.2011.643169

Festl, R., & Quandt, T. (2013). Social relations and cyberbullying: The influence of individual and structural attributes on victimization and perpetration via the Internet. Human Communication Research, 39(1), 101 – 126.

Festl, R., Scharkow, M., & Quandt, T. (2013). Peer influence, internet use and cyberbullying: A comparison of different context effects among German adolescents. Journal of Children and Media, 7(4), 446 – 462.

Haynie, D. L. (2001). Delinquent peers revisited: Does network structure matter? American Journal of Sociology, 106(4), 1013 – 1057. doi:10.1086/320298

Heirman, W., & Walrave, M. (2008). Assessing concerns and issues about the mediation of technology in cyberbullying. Cyberpsychology: Journal of Psychosocial Research on Cyberspace, 2(2), Article 1.

Henry, D. (2001). Classroom context and the development of aggression: The role of normative processes. In F. Columbus (Ed.), Advances in psychology research (Vol. 6, pp. 193 – 213). Huntington, NY: Nova Science.

Hox, J. J. (2010). Multilevel analysis: Techniques and applications (Vol. ,2nd ed.). New York: Routledge.

R. Festl et al. Social Influence on Cyberbullying Involvement

Jaccard, J., Blanton, H., & Dodge, T. (2005). Peer influences on risk behavior: An analysis of the effects of a close friend. Developmental Psychology, 41(1), 135 – 147. doi:10.1037/ 0012-1649.41.1.135

Juvonen, J., & Galvan, A. (2008). Peer influence in involuntary social groups: Lessons from research on bullying. In M. Prinstein & K. Dodge (Eds.), Peer influence processes among youth (pp. 225 – 244). New York: Guilford Press.

Kärnä, A., Voeten, M. J. M., Poskiparta, E., & Salmivalli, C. (2010). Vulnerable children in varying classroom contexts: Bystanders’ behaviors moderate the effect of risk factors on victimization. Merrill-Palmer Quarterly, 56(3), 261 – 282.

Lenhart, A., Purcell, K., Smith, A., & Zickuhr, K. (2010). Social media and mobile internet use among teens and young adults. Pew Internet and American Life Project. Retrieved 30 July, 2013, from http://pewinternet.org/reports/2010/Social-Media-and-Young-Adults.aspx.

Livingstone, S., Haddon, L., Görzig, A., & Olafsson, K. (2011). Risks and safety on the internet: The perspective of European children. Full Findings. London: LSE, EU Kids Online.

Markey, P. M. (2000). Bystander intervention in computer-mediated communication. Computers in Human Behavior, 16(2), 183 – 188.

Mouttapa, M., Valente, T., Gallaher, G., Rohrbach, L. A., & Unger, J. B. (2004). Social network predictors of bullying and victimization. Adolescence, 39(154), 315 – 335.

Neal, J. W. (2010). Social aggression and social position in middle childhood and early adolescence: Burning bridges or building them? The Journal of Early Adolescence, 30(1), 122 – 137. doi:10.1177/0272431609350924

Olweus, D. (1993). Bullying at school: What we know and what we can do. Oxford, England: Blackwell.

Payne, D. C., & Cornwell, B. (2007). Reconsidering peer influences on delinquency: Do less proximate contacts matter? Journal of Quantitative Criminology, 23(2), 127 – 149. doi:10.1007/s10940-006-9022-y

Postmes, T., Spears, R., & Lea, M. (2000). The formation of group norms in computer-mediated communication. Human Communication Research, 26(3), 341 – 371.

Reich, S. M., Subrahmanyam, K., & Espinoza, G. (2012). Friending, IMing, and hanging out face-to-face: overlap in adolescents’ online and offline social networks. Developmental Psychology, 48(2), 356 – 368.

Salmivalli, C., Huttunen, A., & Lagerspetz, K. M. J. (1997). Peer networks and bullying in schools. Scandinavian Journal of Psychology, 38(4), 305 – 312. doi:10.1111/1467-9450. 00040

Salmivalli, C., Kaukiainen, A., & Voeten, M. (2005). Anti-bullying intervention: Implementation and outcome. British Journal of Educational Psychology, 75(3), 465 – 487.

Salmivalli, C., Lagerspetz, K., Björkqvist, K., Österman, K., & Kaukiainen, A. (1996). Bullying as a group process: Participant roles and their relations to social status within the group. Aggressive Behavior, 22(1), 1 – 15. doi:10.1002/(SICI)1098-2337(1996)22:1<1:: AID-AB1>3.0.CO;2-T

Salmivalli, C., & Voeten, M. (2004). Connections between attitudes, group norms, and behaviour in bullying situations. International Journal of Behavioral Development, 28(3), 246 – 258.

Sawyer, A. L., Bradshaw, C. P., & O’Brennan, L. M. (2008). Examining ethnic, gender, and developmental differences in the way children report being a victim of “bullying” on self-report measures. Journal of Adolescent Health, 43(2), 106 – 114. doi:org/10.1016/ j.jadohealth.2007.12.011

Social Influence on Cyberbullying Involvement R. Festl et al.

Sieving, R. E., Perry, C. L., & Williams, C. L. (2000). Change friendships? Examining paths of influence in young adolescents’ alcohol use. Journal of Adolescence Health, 26(1), 27 – 35. doi:10.1016/S1054-139X(99)00056-7

Sijtsema, J. J., Veenstra, R., Lindenberg, S., & Salmivalli, C. (2009). Empirical test of bullies’ status goals: assessing direct goals, aggression, and prestige. Aggressive Behavior, 35(1), 57 – 67. doi:10.1002/ab.20282

Slonje, R., & Smith, P. K. (2008). Cyberbullying: Another main type of bullying? Scandinavian Journal of Psychology, 49(2), 147 – 154. doi:10.1111/j.1467-9450. 2007.00611.x

Slonje, R., Smith, P. K., & Frisen, A. (2013). The nature of cyberbullying, and strategies for prevention. Computers in Human Behavior, 29(1), 26 – 32. doi:10.1016/j.chb.2012.05.024

Smith, P. K. (2012). Cyberbullying and cyber aggression. In S. R. Jimerson, A. B. Nickerson, M. J. Mayer & M. J. Furlong (Eds.), Handbook of school violence and school safety: International research and practice (pp. 93 – 103). New York: Routledge.

Smith, P. K., Mahdavi, J., Carvalho, M., Fisher, S., & Russell, S. (2008). Cyberbullying: Its nature and impact in secondary school students. Journal of Child Psychology and Psychiatry, 49(4), 376 – 385. doi:10.1111/j.1469-7610.2007.01846.x

Spears, R., & Lea, M. (1994). Panacea or panopticon? The hidden power in computer-mediated communication. Communication Research, 21(4), 427 – 459.

Sticca, F., Ruggieri, S., Alsaker, F., & Perren, S. (2013). Longitudinal risk factors for cyberbullying in adolescence. Journal of Community and Applied Social Psychology, 23(1), 52 – 67. doi:10.1002/casp.2136

Tokunaga, R. S. (2010). Following you home from school: A critical review and synthesis of research on cyberbullying victimization. Computers in Human Behavior, 26(3), 277 – 287. doi:10.1016/j.chb.2009.11.014

Vandebosch, H., & van Cleemput, K. (2009). Cyberbullying among youngsters: Profiles of bullies and victims. New Media Society, 11(8), 1349 – 1371. doi:10.1177/146144480 9341263

Walrave, M., & Heirman, W. (2009). Cyberbullying: Predicting victimisation and perpetration. Children and Society, 25(1), 59 – 72. doi:10.1111/j.1099-0860.2009.00260.x

Wasserman, S., & Faust, K. (1994). Social network analysis: Methods and applications. New York: Cambridge University Press.

Xie, H., Swift, D. J., Cairns, B., & Cairns, R. B. (2002). Aggressive behaviors in social interaction and developmental adaptation: A narrative analysis of interpersonal conflicts during early adolescence. Social Development, 11(2), 205 – 224. doi:10.1111/1467-9507. 00195

Ybarra, M. L., & Mitchell, K. J. (2004). Online aggressors, victims, and aggressor/victims: A comparison of associated youth characteristics. Journal of Child Psychology and Psychiatry, 45(7), 1308 – 1316. doi:10.1111/j.1469-7610.2004.00328.x

Ybarra, M. L., & Mitchell, K. J. (2008). How risky are social networking sites? A comparison of places online where youth sexual solicitation and harassment occurs. Pediatrics, 121(2), 350 – 357.

Copyright of Human Communication Research is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.