Effects of Aggressive Traits On Cyberbullying: Mediated Moderation or Moderated Mediation?

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/331791150

Effects of aggressive traits on cyberbullying: Mediated moderation or


moderated mediation?

Article  in  Computers in Human Behavior · August 2019


DOI: 10.1016/j.chb.2019.03.015

CITATIONS READS

10 700

6 authors, including:

Shen Liu Lin Zhang


University of Science and Technology of China Ningbo University
51 PUBLICATIONS   139 CITATIONS    47 PUBLICATIONS   104 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Shen Liu on 25 March 2019.

The user has requested enhancement of the downloaded file.


Computers in Human Behavior 97 (2019) 167–178

Contents lists available at ScienceDirect

Computers in Human Behavior


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

Full length article

Effects of aggressive traits on cyberbullying: Mediated moderation or T


moderated mediation?
Minghua Songa,b,1, Zhuan Zhua,1, Shen Liuc,∗,1, Hang Fana, Tingting Zhua, Lin Zhanga,∗∗
a
Department and Institute of Psychology, Ningbo University, Ningbo, Zhejiang, China
b
Mental Health Education Guidance Center, Huzhou University, Huzhou, Zhejiang, China
c
School of Humanities and Social Sciences, University of Science and Technology of China, Hefei, Anhui, China

A R T I C LE I N FO A B S T R A C T

Keywords: To explore the process of cyberbullying, the current study investigated the relationships among aggressive traits,
Chinese college students beliefs about aggression, network public opinion, and cyberbullying. A questionnaire survey was conducted
Aggressive traits among 693 Chinese college students. Based on the findings, we constructed two competing models (mediated
Cyberbullying moderation and moderated mediation). The results revealed the following: (1) Aggressive traits and beliefs about
Network public opinion
aggression were all positively correlated with cyberbullying, aggressive traits were positively correlated with
Beliefs about aggression
beliefs about aggression, and aggressive traits were negatively correlated with network public opinion. (2) The
Mediated moderation model
effect of aggressive traits on beliefs about aggression and cyberbullying was moderated by network public
opinion. When network public opinion included irrational comments, aggressive traits directly and indirectly
affected cyberbullying. When network public opinion included rational comments, aggressive traits only directly
affected cyberbullying. Finally, (3) the effect of aggressive traits on cyberbullying was moderated by network
public opinion and the moderating effect of network public opinion was partly realized through the mediating
effect of beliefs about aggression. The findings contribute to our understanding of when and how aggressive
traits affect cyberbullying. We found support for the mediated moderation model, and the results have im-
plications for prevention and intervention in cyberbullying.

1. Introduction the consideration of the comprehensive mechanism of cyberbullying


from individual and environmental factors based on the in-
Cyberbullying refers to applying information and communication dividual–environmental interaction model (Zheng et al., 2017). In ad-
technology to engage in deliberate, repetitive, or hostile behaviors to- dition, previous studies showed that people's views, attitudes, and
wards others (Jenaro, Flores, & Frías, 2017; Kowalski, Giumetti, opinions about an event on the Internet platform can influence cyber-
Schroeder, & Lattanner, 2014). Cyberbullying not only severely affects bullying (Li, 2006; Çetin, Yaman, & Peker, 2011). However, no study
the normative order of the network platform but also is closely related has yet investigated the effect of Internet public opinion on cyberbul-
to a variety of adverse consequences such as low self-esteem, depres- lying. Based on the individual–environment interaction model, the
sion, and even suicidal ideation and behavior (Gini & Espelage, 2014; current study comprehensively investigated the effects of individual
Holt et al., 2015; Reed, Cooper, Nugent, & Russell, 2016; Savage & (aggressive traits and beliefs about aggression) and environmental
Tokunaga, 2017; Selkie, Kota, Chan, & Moreno, 2015). Although ex- factors (network public opinion) on cyberbullying to provide targeted
isting studies have paid close attention to the mechanism of cyberbul- suggestions for effective prevention and reduction of cyberbullying.
lying, such as the effects on cyberbullying of narcissistic, normative
belief in aggressive behaviors; empathy; and gender differences (Li, Li, 1.1. The effect of aggressive traits on cyberbullying
& Zhang, 2016; Rey et al., 2016; Topcu & Erdurbaker, 2012), these
studies were all conducted from a certain perspective. Thus, they lack Cyberbullying is defined as certain aggressive behaviors that occur

∗∗
Corresponding author. Department and Institute of Psychology, Ningbo University, No. 616 Fenghua Road, Jiangbei District, Ningbo, Zhejiang, 315211, China.

Corresponding author. School of Humanities and Social Science, University of Science and Technology of China, No. 96 Jinzhai Road, Shushan District, Hefei,
Anhui, 230022, China.
E-mail addresses: liushenpsy@ustc.edu.cn (S. Liu), zhanglin1@nbu.edu.cn (L. Zhang).
1
Minghua Song, Zhuan Zhu and Shen Liu Contributed equally to this study and share the first authorships.

https://doi.org/10.1016/j.chb.2019.03.015
Received 14 November 2018; Received in revised form 10 March 2019; Accepted 15 March 2019
Available online 20 March 2019
0747-5632/ © 2019 Published by Elsevier Ltd.
M. Song, et al. Computers in Human Behavior 97 (2019) 167–178

in cyberspace, specifically malicious, repetitive, and hostile behavior 1.2. The mediated moderation model
towards others by means of electronic information media, such as
spreading rumors, backstabbing, spreading personal information, and According to the trait activation theory, whether traits (aggressive
so on (Antoniadou & Kokkinos, 2015). Cyberbullying is a derivative traits) can predict behavior depends on whether trait-related cues are
form of traditional bullying and a new form of aggression with the provided in the context. If there is a trait-related cue, the trait is acti-
development of the Internet (Baroncelli & Ciucci, 2014; Hutson, 2016). vated by the situation, and individuals are more influenced by the trait
In traditional bullying, the bully deliberately causes physical and psy- to maintain the trait-related behaviors. On the contrary, if there is no
chological harm to the other party directly or indirectly. It is a con- trait-related cue, the trait cannot be activated by the situation, and the
scious act usually manifested as verbal provocation (such as insult, individuals' behaviors are more affected by the situation (Tett &
defamation, exclusion, intimidation) and physical harm (Li et al., Burnett, 2003). That is, the effects of aggressive traits on cyberbullying
2016). Bullying generally consists of three elements, including inten- are moderated by situational factors. The occurrence of cyberbullying
tional harmful acts, repeated occurrences over time, and imbalance of cannot be separated from the main carrier of the Internet. As a new
power between the two sides (Kowalski et al., 2014). However, cy- mass medium, the convenience, anonymity, and interactivity of the
berbullying has subverted these traditional elements of bullying and network platform makes it popular among netizens. When netizens are
makes it unnecessary for both sides to meet face to face. Therefore, willing to use the network platform to express their views, opinions,
bullying no longer requires the “power principle” (Li et al., 2016). With and comments, the network public opinion comes into being (Jiang &
the help of new technologies, cyberbullying permits anonymity and fast Liu, 2011). Network public opinion, as a new form of public opinion
communication, and has no space- or time-based constraints. Compared extended by traditional public opinion in the network, refers to the
to traditional bullying, the forms of cyberbullying are diverse. They can opinion or speech about which the public (especially netizens) shows a
be divided into emotional flaming, online harassment, online hate, certain influence, a tendency of opinion or speech regarding some
cyberstalking, denigration, masquerade, outing, and online exclusion public events through network language or other means, taking the
(Athanasiades, Baldry, Kamariotis, Kostouli, & Psalti, 2016; Keipi et al., network as the platform (Hu, 2016). Although the Internet provides
2016). According to the general aggression model, individual and en- netizens with convenience, there are many irrational and rational ele-
vironmental factors affect individuals' internal cognitive process, sub- ments involved in the network public opinion (Jiang & Liu, 2011). The
sequent evaluation, and decision-making process to further influence irrational expression of network public opinion refers to the fact that
the occurrence of aggression (Montuoro & Mainhard, 2017). Individual netizens can easily make judgments on a given event without rigorous
factors include individual personality traits, beliefs, and gender. For consideration and only based on their subjective judgment and pre-
example, Liu, Jiang, Ren, Li, and Xu (2015) found that trait anger, as an ference, which is often characterized by emotionality, extremes, and
input variable, activated the intrinsic level of hostility cognition to in- negative tropism (Zhang & Yan, 2016). Previous studies have shown
crease the possibility of aggression. Zheng et al. (2017) found that in- that the irrational expression atmosphere of network public opinion
dividual normative beliefs about aggression affected moral disengage- could lead to “human flesh search,” spreading rumors, abuse, and
ment and Internet morality to indirectly influence cyberbullying. Topcu malicious defamation of others (Liu, 2014a, 2014b; Çetin et al., 2011).
and Erdurbaker (2012) found that differences in gender affected in- Näsi et al. found that there was a significant association between ex-
dividual emotional and cognitive empathy to indirectly influence cyber- posure to websites relating to eating disorders and online harassment.
and traditional bullying. Aggressive traits, as stable personality traits Hawdon et al. (2016) found that there is a considerable amount of hate
characterized by hostile cognition, anger, and readiness, play an im- material online, but the degree to which individuals from different
portant role in violent and problematic behaviors (Geel, Goemans, countries are exposed to these materials varies, and thus there can be
Toprak, & Vedder, 2017; Molapour, Lindström, & Olsson, 2016). For country-wide differences in degree of online hate. In contrast, the ra-
example, individuals with aggressive traits are more likely to make tional expression of network public opinion can lead individuals to
hostile attributions, which increases the likelihood of anger and ag- correctly view problems as well as events and hold appropriate atti-
gressive behaviors (Gagnon et al., 2017). Smith et al. (2011) found that tudes as well as beliefs about the problem or event, thus causing less
individuals high in aggressive traits were less sensitive to aggressive cyberbullying (Xing, 2015). In addition, the individual–environment
behaviors than were those low in aggressive traits and were more likely interaction model also points out that individuals' behaviors are formed
to regard a certain behavior as non-aggressive, leading to more ag- and developed in the interaction between individuals and environment.
gressive behaviors among individuals high in aggressive traits (Gagnon Individuals’ own factors also interact with the environmental factors to
et al., 2017). In addition, both proactive and reactive aggression are influence individual development (Peter & Petermann, 2018). In-
positively correlated with cyberbullying (Ang, Huan, & Florell, 2014; dividuals carry out aggressive behaviors under the influence of en-
Savage & Tokunaga, 2017). Compared to individuals low in aggressive vironmental factors, and different network atmospheres of public opi-
traits, individuals high in aggressive traits were more likely to interpret nion expression can moderate the relationship between aggressive traits
others’ behaviors as hostile or malicious and tended to have hostile and cyberbullying. For example, the irrational expression of online
cognitive bias towards others, resulting in increased hostility and ag- public opinion is the result of group comments, and the individuals with
gression. By virtue of the anonymity and convenience of the network, aggressive traits in this situation will show more cyberbullying due to
they arbitrarily released the negative emotions that could not be dis- the spread of responsibility (Long, 2014). The rational expression of
pelled in real life to other individuals in the network environment and network public opinion will reduce or even change the negative judg-
then engaged in cyberbullying (Tremblay & Belchevski, 2004). Studies ment of aggressive individuals regarding the current situation and
have shown that general bullying can predict cyberbullying, and most generate less cyberbullying (Dewall, Anderson, & Bushman, 2011).
cyberbullies are also more likely to attack others in real life (Hemphill Therefore, we proposed the following Hypothesis:
et al., 2012; Slonje & Smith, 2008). Thus, individuals with aggressive
Hypothesis 2. Network public opinion will moderate the relationship
traits in real life may also show more aggression on the network plat-
between aggressive traits and cyberbullying.
form, causing them to bully others online. In other words, cyberbul-
lying is a displacement of individuals with aggressive traits in cyber- According to the general aggression and aggressive social cognition
space to express aggressive behaviors in real life. Therefore, we models, individual factors seldomly have direct influence on in-
proposed the following Hypothesis: dividuals' behaviors but often are influential through individual in-
ternal cognitive processes. As an individual trait factor, aggressive traits
Hypothesis 1. Aggressive traits will be positively correlated with
often rely on internal cognitive processes to evoke aggressive behaviors.
cyberbullying.
Previous studies found that individuals with aggressive traits

168
M. Song, et al. Computers in Human Behavior 97 (2019) 167–178

constructed a schema of aggressive knowledge based on their accu- 1.3. The moderated mediation model
mulated life experience, which could affect their aggressive cognition
(Dewall et al., 2011) and, thus, more easily form beliefs about aggres- As mentioned above, according to the general aggression model, the
sion. Belief about aggression refers to individuals' perception of whe- mechanism of aggressive behaviors is to activate the individual intrinsic
ther aggression is acceptable or not when settling disputes and ex- information processing model by taking individual and situational
pressing hatred (Zhang, Liu, Xu, Wu, & Yang, 2017; Zhen, Xie, Zhang, factors as input variables (such as aggressive traits and network public
Wang, & Li, 2011) and is one of the main incentives for aggressive opinion); then, the individual internal information processing model
behaviors (Montuoro & Mainhard, 2017). Previous studies found that cognizes the input variables. Afterwards, the aggressive schema (such
aggressive traits could positively predict beliefs about aggression and as beliefs about aggression) is activated, and the individual is guided by
that individuals with higher aggression were more likely to recognize the aggressive schema to initiate aggressive behaviors (Dewall et al.,
and accept aggressive behaviors as a means of settling disputes and 2011; Montuoro & Mainhard, 2017). This theory particularly empha-
expressing hatred (Adams & Ireland, 2017; Geel et al., 2017). Similarly, sizes the dominant role of aggressive traits in aggressive behaviors. On
individuals' beliefs about aggression reflect their recognition and ac- one hand, aggressive traits increase hostile cognition, generate hostile
ceptance of aggressive behaviors and have a close correlation with interpretation, and activate aggressive schema. On the other hand, they
aggressive behaviors (Montuoro & Mainhard, 2017; Zhang et al., 2017). interfere with the construction of aggressive knowledge schema, affect
Some studies found that beliefs about aggression could significantly beliefs about aggression, and then lead to aggressive behaviors (Adams
affect people's perception of aggressive cues, and higher beliefs about & Ireland, 2017; Ang, 2015; Wright & Li, 2013). Belief about aggression
aggression could make them inclined to regard aggressive behaviors as is a belief used to evaluate the acceptability of behaviors. Individuals
a way of coping for social recognition or acceptance (Maier & James, make aggressive or non-aggressive judgments of others' behavioral in-
2014). Other studies also found that the cognitive structure (beliefs tention based on their own normative beliefs about aggression
about aggression) of individuals regarding aggressive behavior posi- (Huesmann & Kirwil, 2007). It can effectively predict the occurrence
tively predicted traditional bullying (Ang, 2015; Williams & Guerra, probability of individuals' aggressive behaviors with different norma-
2007; Wright & Li, 2013; Zheng et al., 2017). Cyberbullying is a special tive beliefs, and the normative beliefs holding supportive aggressive
derivative of traditional aggressive behaviors (Baroncelli & Ciucci, behaviors are positively correlated with actual aggressive behaviors
2014); individuals with higher beliefs about aggression may also per- (Wright & Li, 2013). Wilkowski and Robinson (2008) put forward the
form more cyberbullying. Therefore, we proposed the following Hy- Integrative Cognitive Model based on integrating existing theoretical
pothesis: models and empirical evidence. They believed that different personality
traits affected the internal cognition of individuals and led to different
Hypothesis 3. Beliefs about aggression will mediate the relationship
behavioral outcomes. This suggests that there is a direct correlation
between aggressive traits and cyberbullying.
between individuals' beliefs about aggression and cyberbullying, and
However, different atmospheres of network public opinion expres- the influence of individual trait factors on cyberbullying may need to be
sion may play a specific and complex role in multiple pathways related exerted through beliefs about aggression. For example, the joint effect
to cyberbullying. For example, Zhang et al. (2017) found that virtual of exclusion and beliefs about aggression can further strengthen in-
situations of repeated exposure to violent information tended to acti- dividuals’ aggressive behaviors (Poon & Chen, 2014). Beliefs about
vate and strengthen individuals’ aggressive schemas, namely, to form aggression partially mediated the relationship between narcissistic
automatic links among memory, emotion, and aggressive behaviors, personality traits and cyberbullying (Ang, Tan, & Talib, 2011).
thus prompting individuals to form aggressive personality traits, gra- As an important individual difference, the effect of aggressive traits
dually influencing or changing adulthood individual cognitive belief on cyberbullying is likely to be moderated by environmental factors.
systems of aggressive behavior, and then triggering cyberbullying. For example, parental control, a family environment variable, moder-
Compared to the rational expression atmosphere of network public ates the relationship between trait self-esteem and cyberbullying
opinion, when individuals with aggressive traits are exposed to the (Palermiti, Servidio, Bartolo, & Costabile, 2017). As another environ-
environment information of the irrational expression of network public mental variable, network public opinion has a certain effect on in-
opinion, their internal aggressive beliefs are also improved, and the dividuals' thinking and behaviors (Liu et al., 2015). In other words,
probability of bullying others on the network increases accordingly network public opinion may moderate the effect of beliefs about ag-
(Dewall et al., 2011; Kowalski et al., 2014). Therefore, we proposed the gression on cyberbullying. According to the self-regulatory executive
following Hypothesis: function model (Paananen et al., 2019), individuals with aggressive
traits are unable to calm down and easily generate violence cognition
Hypothesis 4(a). Beliefs about aggression will mediate the interactive
and mood, resulting in the failure of individual self-regulatory execu-
effects of aggressive traits and network public opinion on
tive function. Surrounded by the atmosphere of network public opinion
cyberbullying.
irrational expression, individuals' beliefs about aggression will be en-
Thus, the current study constructed a mediated moderation model hanced, increasing the level of impulse, decreasing the ability of self-
for the relationship between aggressive trait and cyberbullying based control, and making the individual unable to reasonably estimate the
on H1, H2, H3, and H4a (see Fig. 1). consequences of the behavior or correctly and reasonably formulate the
coping strategies of behavior. Thus, these individuals very easily show
cyberbullying when using the network. Compared to the traditional
environment of public opinion, the network environment provides a
new space for expression and discussion for the formation of public
opinion as well as a relatively free communication atmosphere for
public discussion. It can directly reflect the public's views or opinions
on certain public events in cyberspace (Zhang & Yan, 2016). The ex-
pression of network public opinion includes rational and irrational
elements. Compared to the rational expression of network public opi-
nion, the irrational expression of network public opinion will affect
individuals' attitudes and beliefs about a certain event and then more
easily cause cyberbullying (Xing, 2015; Çetin et al., 2011). According to
Fig. 1. The mediated moderation model. the cognitive connection theory, individuals with high beliefs about

169
M. Song, et al. Computers in Human Behavior 97 (2019) 167–178

aggression influenced by aggressive words in online platforms will be when the mediator variable is added, the mediating effect is more ap-
more inclined to recognize and accept aggression to settle disputes and parent. When reduced to 0, the regression coefficient is called the full-
express hatred in the irrational expression atmosphere of network mediator (Preacher & Hayes, 2008). As mentioned earlier, if the re-
public opinion, thus promoting the occurrence of cyberbullying. While lationship between aggressive traits as the independent variable and
individuals influenced by the group atmosphere will hold appropriate cyberbullying as the dependent variable is affected by the third variable
attitudes and beliefs about an event and make rational judgments in the of network public opinion, then the network public opinion is a mod-
rational expression atmosphere of network public opinion, they cannot erating variable, which affects the direction (positive or negative) and
easily activate beliefs about aggression, which will reduce the occur- the strength (strong or weak) of the relationship between aggressive
rence of cyberbullying (Liu et al., 2015). It can be seen that different traits and cyberbullying. For example, the relationship between “ag-
expression atmospheres of network public opinion affect individuals' gressive traits” and “cyberbullying” is influenced by “network public
cognitive belief systems, thus causing the increase or decrease of cy- opinion” in the current study. When in an irrational expression atmo-
berbullying (Peter & Petermann, 2018). Specifically, when in the irra- sphere of network public opinion, aggressive traits have a positive
tional expression atmosphere of network public opinion, with the in- correlation with cyberbullying; while in a rational expression atmo-
crease of individuals' beliefs about aggression, cyberbullying is sphere of network public opinion, the correlation between aggressive
increased; while in the rational expression atmosphere of network traits and cyberbullying is not significant. The purpose of the moder-
public opinion, with the decrease of individuals' beliefs about aggres- ating effect analysis is to explore when aggressive traits affect cyber-
sion, cyberbullying is also decreased. Therefore, we proposed the fol- bullying or when they have a significant impact (Muller et al., 2005). If
lowing Hypothesis: the relationship between aggressive traits as the independent variable
and cyberbullying as the dependent variable is affected by the third
Hypothesis 4(b). Network public opinion will moderate the
variable of beliefs about aggression, then beliefs about aggression are a
relationship between beliefs about aggression and cyberbullying.
mediating variable, representing a mechanism in which aggressive
Thus, the current study constructed a moderated mediation model traits indirectly affect cyberbullying by influencing beliefs about ag-
for the relationship between aggressive trait and cyberbullying based gression. The purpose of mediation effect analysis is to explore how
on H1, H2, H3, and H4b (see Fig. 2). aggressive traits as the independent variable affect cyberbullying as the
dependent variable (Muller et al., 2005). If a model contains more than
1.4. Competing models: a mediated moderation model or a moderated three variables as well as both moderation and mediation variables,
mediation model? these variables in the different position and role of model can produce
different models. The mediated moderation model and the moderated
In the study of psychology and behavior, the relationship between mediation model contain two kinds of common models of moderation
the predictor and dependent variables is often influenced by a “third and mediation variables. Among them, the mediated moderation model
variable,” namely, a moderation variable and mediation variable implies that the effect of the independent variable on dependent vari-
(Muller, Judd, & Yzerbyt, 2005; Ye & Wen, 2013). Both moderators and able is influenced by the moderator variable while the moderator effect
mediators can explain the relationship between predictor and depen- (at least partially) is affected by the mediating variable (Baron & Kenny,
dent variables, though there are differences between them. If the re- 1986; Wen et al., 2012; Ye & Wen, 2013). That is to say, the mediated
lationship between the predictor variable and the dependent variable is moderation model indicates that the effect of aggressive traits on cy-
a function of the variable M, then M is called the moderator variable. berbullying is influenced by network public opinion, and the moder-
Specifically, the moderator can influence the direction (positive or ating effect (at least in part) works through beliefs about aggression.
negative) and intensity (strong or weak) of the relationship between the The moderated mediation model means that the independent variable
predictor and dependent variables. The moderator can be qualitative influences the dependent variable through the mediating variable, and
(such as gender or race) or quantitative (such as age or length of edu- the mediation process is regulated by the moderating variable (Baron &
cation). The moderating effect is often considered when the relation- Kenny, 1986; Wen et al., 2006; Ye&Wen, 2013). In other words, the
ship between the predictor and dependent variables is slightly stronger moderated mediation model indicates that aggressive traits impact
or weaker or changes direction. However, the meaning of the mediation cyberbullying through beliefs about aggression, and this mediation
variable is different from that. If the predictor variable influences the process is moderated by network public opinion. However, the pur-
dependent variable through the variable M, then M is called the med- poses and focuses of these two models are quite different, as are their
iator variable. The mediator variable plays an indirect role in ex- arguments and explanations. The focus of the mediated moderation
plaining how the predictor variable affects the dependent variable model is to consider whether the direction (positive or negative) and
through it. The relationships among these three variables are examined strength (strong or weak) of the relationship between aggression traits
when examining the mediating effect. Firstly, it is assumed that there is as the independent variable and cyberbullying as the dependent vari-
a significant correlation between the predictor and dependent vari- able is affected by the network public opinion as the moderating vari-
ables, and there is a significant correlation between the predictor and able, namely, the moderating effect. Secondly, it considers how the
mediator variables. If the correlation between the predictor and de- network public opinion of moderating variable works—that is, whether
pendent variables or the regression coefficient is significantly decreased it works through beliefs about aggression as the mediating variable. The
focus of the moderated mediation model is to consider the effect of the
mechanism of aggression traits as the independent variable on cyber-
bullying as the dependent variable, namely, the mediation effect and,
secondly, whether the mediation process is moderated—that is, when
the mediation effect is stronger and when it is weaker ( HYPERLINK \l
"bib62" \o "bib62" Ye&Wen, 2013).
The mediated moderation model can further explore the relation-
ship between aggressive traits and cyberbullying and its mechanism,
which means aggressive traits are one of the important risk factors for
cyberbullying, preliminarily illustrates the conditions under which ag-
gressive traits work, and also reveals how aggressive traits affect cy-
berbullying under different conditions. The moderated mediation
Fig. 2. The moderated mediation model. model is more effective in explaining the phenomenon of cyberbullying

170
M. Song, et al. Computers in Human Behavior 97 (2019) 167–178

than the pure moderation or mediation model. The mediated modera- comment on post events differently. That is, most of the comments were
tion model is designed to explain how aggressive traits and network supportive of a star's behavior in the irrational expression group,
public opinion interactively affect cyberbullying and whether beliefs whereas while some of the comments were supportive of his or her
about aggression mediate their interactive effect. The moderated behaviors in the rational expression group, others were opposed, and a
mediation model can further explore how aggressive traits affect cy- few were neutral. Then, the participants were asked to evaluate the
berbullying (the mediation mechanism) and under what conditions (the attitudes of the parties according to the post events, and three different
moderation mechanism). The mediation mechanism can answer how options were set for the participants to choose according to their true
aggressive traits as the independent variable affect cyberbullying as the attitudes. The three attitudinal evaluation options were (1) opposing
dependent variable, but it only focuses on the “process” and “com- the violence of the parties, with a score of 1; (2) supporting the violence
monness” of the occurrence of variable relations. It cannot answer who of the parties, with a score of 2; and (3) maintaining a neutral attitude,
is more significant in terms of the effect of aggressive traits as the in- with both parties responsible, and with a score of 3. To test the effec-
dependent variable on cyberbullying as the dependent variable—that tiveness of the manipulation, an independent sample t-test was con-
is, the “condition” and “personality” issues of the occurrence of variable ducted for the rational and irrational comment groups. There was a
relations. As far as the current study is concerned, the mediating effect significant difference between the rational (M = 2.75, SD = 0.50) and
of beliefs about aggression can only explain that it is the proximal factor irrational comment groups (M = 2.28, SD = 0.91; t(690) = 4.70,
between aggressive traits and cyberbullying; namely, beliefs about ag- p < 0.001, d = 0.64). This indicated that the material manipulation of
gression mediate the effects of aggressive traits on cyberbullying. the current study was effective, and different expressions of network
However, the indirect effect of beliefs about aggression may be mod- public opinion ere formed through different comment tendencies of the
erated by other factors. That is, the indirect effect is more significant for virtual community page posts.
an individual in a certain situation and may not be obvious for an in-
dividual in another situation. The moderated mediation model is de- 2.2.2. Aggressive traits scale
signed to investigate the mediating effect of beliefs about aggression The Aggression Questionnaire was compiled by Buss and Perry
between aggressive traits and cyberbullying, and this mediation effect (1992) and revised by Zhang et al. (2017). It is a 29-item questionnaire
is moderated by network public opinion. that uses a 5-point Likert scale ranging from 1 (not at all) to 5 (full) and
Based on the abovementioned theories and empirical evidences, the includes four factors: physical aggression, verbal aggression, anger, and
current study comprehensively investigated the effects of these vari- hostility. Responses across the 29 items were summed up with higher
ables of network public opinion, aggressive traits, and beliefs about scores indicating higher levels of aggressive traits. In the current study,
aggression on cyberbullying and put forward five hypotheses, which the second-order CFA model generated a very good fit with χ2/
were intended to test the two competitive models of mediated mod- df = 3.51, CFI = 0.96, NFI = 0.94, GFI = 0.93, RMSEA = 0.002, and
eration and moderated mediation according to the general aggression both the absolute and value-added adaptation indexes were within the
model and the trait activation theory. In addition, the current study acceptable range. In the current study, the Cronbach's alpha coefficient
examined the roles of beliefs about aggression and network public was 0.86.
opinion between aggressive traits and cyberbullying, which is con-
ducive to in-depth understanding of the effect mechanism of aggressive 2.2.3. Beliefs about Aggression Scale
traits on cyberbullying and the conditions of its occurrence, to provide The Normative Beliefs about Aggression Scale was compiled by
effective theoretical basis for the prevention of and intervention in Huesmann and Guerra (1997) and revised by Zhang et al. (2017). It is a
cyberbullying. 20-item questionnaire that uses a 4-point Likert scale ranging from 1
(completely unreasonable) to 4 (completely reasonable); 12 items are
2. Methods retaliatory beliefs about aggression in specific situations, and the re-
maining 8 items are beliefs about aggression in general situations. Re-
2.1. Participants sponses across the 20 items were summed with higher scores indicating
higher levels of beliefs about aggression. In the current study, the
In the current study, 700 questionnaires were randomly assigned to second-order CFA model generated a very good fit with χ2/df = 3.02,
college students from three universities in China via convenience CFI = 0.94, NFI = 0.93, GFI = 0.94, RMSEA = 0.031, and both the
sampling. During the process, seven of them failed to answer some absolute and value-added adaptation indexes were within the accep-
questions and were excluded. Thus, there were 693 valid questionnaires table range. In the current study, the Cronbach's alpha coefficient was
(211 males and 482 females), and the effective recovery rate was 99%. 0.87.
Internet age ranged from 1 to 23 years (M = 7.65 years, SD = 2.24
years). Time spent online ranged from 0.5 to 24.0 h per day with an 2.2.4. Cyberbullying scale
average of 5.41 h per day (SD = 2.52 h). The current study followed the The Cyber Victim and Bullying Scale was compiled by Çetin et al.
principle of voluntariness, and course credits were given to participants (2011). As this study aimed to measure bullying of others, irrelevant
after completion. The current study was approved by the Ethics items in the scale were eliminated, and the translated scale was revised
Committee of the authors’ university in accordance with the ethical to form the Chinese revision scale. The scale is a 25-item questionnaire
principles of the Declaration of Helsinki. All participants gave written that uses a 5-point Likert scale ranging from 1 (never) to 5 (always).
informed consent in accordance with the ethical principles of the Responses across the 25 items were summed up with higher scores
Declaration of Helsinki. indicating higher levels of cyberbullying. In the current study, the
second-order CFA model generated a very good fit with χ2/df = 2.58,
2.2. Measures CFI = 0.95, NFI = 0.94, GFI = 0.95, RMSEA = 0.020, and both the
absolute and value-added adaptation indexes were within the accep-
2.2.1. Manipulation material of network public opinion guidance table range. In the current study, the Cronbach's alpha coefficient was
According to Huang, Ding, Wei, and Hong (2018), the manipulation 0.97.
of materials of network public opinion orientation (rational or irra-
tional expression) were presented in the form of virtual community 2.3. Procedure
page posts. The information presented includes post events and com-
ments from netizens on post events, where the post events are the same The current study was conducted by well-trained graduate students
(e.g., a star was reported to have beaten a reporter), but netizens tend to majoring in psychology. With the help and cooperation of school

171
M. Song, et al. Computers in Human Behavior 97 (2019) 167–178

teachers, the group testing was carried out with the class as a unit with provided in Table 1. The results of the bivariate correlations showed
unified procedures. Participants first filled in the demographic variables that aggressive traits were positively correlated with beliefs about ag-
(such as gender, grade, network age), and then were randomly pre- gression and cyberbullying (p < 0.01), and beliefs about aggression
sented with virtual community page posts with rational or irrational were positively correlated with cyberbullying (p < 0.01). Network
expression of network public opinion. They were asked to evaluate the public opinion was not significantly correlated with belief about ag-
event of the post and the evaluation results were neither good nor bad. gression and cyberbullying except for a negative correlation with ag-
Then, the participants completed the measurement of aggressive traits, gressive traits. In addition, Internet age, grade, and Internet time were
belief about aggression, and cyberbullying, and finally, the experi- all significantly positively correlated with cyberbullying. Therefore,
menter collected questionnaires on the spot. Internet age, grade, and Internet time were used as control variables in
subsequent analysis to explore independent effect of aggressive traits
2.4. Statistical analyses and beliefs about aggression on cyberbullying.

Data collected in the current study were recorded on a computer


and processed using SPSS 22.0 and Amos 21.0. According to Erceg- 3.3. The relationship between aggressive traits and cyberbullying: a mixed
Hurn and Mirosevich (2008), the bootstrap confidence intervals were model analysis
used to test the significance of the regression coefficients. This method
does not need to assume normality of the data but reconstructs the According to Ye and Wen (2013), if the mediated moderation model
sample distribution by adopting a random sampling method based on satisfied the following three requirements, it indicated the existence of
the original sample (a total of 5000 resamples in the current study, each mediated moderation effect (see Table 2). In the first step, aggressive
with a sample size of 693), which then obtains a robust standard error traits as the independent variable had a significant effect on cyberbul-
of the parameter estimation and confidence intervals with 95% bias lying as the dependent variable (β = 0.47, p < 0.001). Aggressive
correction. If the confidence interval does not contain zero, it indicates traits as the independent variable and network public opinion as the
statistical significance. moderator should significantly interact to influence cyberbullying as
the dependent variable (β = −0.24, p < 0.001). Next, as a second
step, aggressive traits as the independent variable should have a sig-
3. Results nificant effect on beliefs about aggression as the mediator (β = 0.25,
p < 0.001), and the interaction effect of aggressive traits as the in-
3.1. Common method biases test dependent variable × network public opinion as the moderator on be-
liefs about aggression as the mediator was required to be significant
The common method bias (CMB) may exist in the current study as (β = −0.28, p < 0.001). Finally, in the third step, beliefs about ag-
all of the questions in the survey were answered by the same re- gression as the mediator was added into the model while additionally
spondent. We used two techniques to check whether there was a threat controlling for the interaction of beliefs about aggression as the med-
of CMB. First, Harman's single-factor test was conducted to determine if iator × network public opinion as the moderator. The third condition to
the variance of our data came largely from a common method source support a mediated moderation model required that beliefs about ag-
(Podsakoff, Mackenzie, Lee, & Podsakoff, 2003). We subjected all of the gression as the mediator were significantly related to cyberbullying as
measurement items of major constructs and control variables to a the dependent variable (β = 0.12, p < 0.05) and that the relationship
principal component analysis to see how many orthogonal components between cyberbullying as the dependent variable and the interaction of
would emerge. Fourteen components with eigenvalues of more than 1.0 aggressive traits as the independent variable × network public opinion
emerged, and the largest single component accounted for 20.38% of the as the moderator was reduced in size (β = −0.18, p < 0.01) after
variance. Secondly, the single-factor model was constructed with all controlling for the interaction of beliefs about aggression as the med-
dimensions of aggressive traits, beliefs about aggression, and cyber- iator × network public opinion as the moderator. Predictors were
bullying as indicators. The results indicated that the CFA model gen- standardized in each equation, and the network age, grade, and net-
erated a poor fit with χ2/df = 17.182, RMSEA = 0.157, NFI = 0.186, work time were controlled. The variance expansion factor of all pre-
CFI = 0.743, SRMR = 0.094, indicating that there was no significant dictive variables was less than 1, so there was no multicollinearity
CMB between the variables. problem. The results of further test on the integrated model indicated
that the degree of fit of the model was good (χ2/df = 3.02, CFI = 0.92,
3.2. Descriptive statistics and correlation analysis NFI = 0.95, GFI = 0.94, RMSEA = 0.032; see Fig. 3). Among them, the
interaction between aggressive traits and network public opinion sig-
Descriptive statistics and the correlation matrix of aggressive traits, nificantly negatively predicted cyberbullying (γ = −0.23, p < 0.001),
beliefs about aggression, network public opinion, and cyberbullying are the interaction between aggressive traits and network public opinion

Table 1
Descriptive statistics and correlation matrix of all variables.
Variables M SD 1 2 3 4 5 6 7 8

1 Sex a
– – 1
2 Grade – – −0.07 1
3 Internet age 5.83 2.70 0.10 0.16∗∗ 1
4 Internet time 4.00 3.01 0.20∗∗ 0.40∗∗ 0.27∗∗ 1
5 Aggressive traits 73.27 15.07 −0.04 0.09 0.04 0.21∗∗ 1
6 Belief about aggression 43.89 4.85 0.09 0.03 0.06 0.03 0.26∗∗ 1
7 Network public opinion b
– – −0.02 0.33∗∗ 0.10 0.06 −0.13∗ −0.08 1
8 Cyberbullying 36.51 15.57 0.10 0.28∗∗ 0.18∗∗ 0.29∗∗ 0.52∗∗ 0.34∗∗ −0.07 1

Note. N = 693.

p-value < 0.05, ∗∗p-value < 0.01.
a
Gender is a dummy variable, female = 0, male = 1.
b
Network public opinion is a dummy variable, irrational comment = 0, rational comment = 1.

172
M. Song, et al. Computers in Human Behavior 97 (2019) 167–178

Table 2
Testing the mediated moderator effect model.
Variables Model 1(Y: Cyberbullying) Model 2 (M: Beliefs about aggression) Model 3(Y: Cyberbullying)

B SE β 95%CI B SE β 95%CI B SE β 95%CI

∗∗∗ ∗∗∗ ∗∗∗


X 0.47 0.53 0.47 [0.37, 0.58] 0.25 0.07 0.25 [0.11, 0.39] 0.44 0.05 0.44 [0.34, 0.55]
U −0.17 0.05 −0.17∗∗ [−0.27, −0.07] −0.09 0.07 −0.09 [−0.22, 0.04] −0.16 0.05 −0.16∗∗ [−0.26, −0.06]
X×U −0.24 0.05 −0.24∗∗∗ [−0.34, −0.14] −0.28 0.07 −0.28∗∗∗ [−0.41, −0.15] −0.18 0.05 −0.18∗∗ [−0.28, −0.08]
M 0.12 0.05 0.12∗ [0.01, 0.22]
M×U −0.12 0.05 −0.12∗ [−0.22, −0.01]
Sex 0.16 0.10 0.16 [−0.03, 0.36] 0.17 0.13 0.17 [−0.09, 0.43] 0.14 0.01 0.14 [−0.05, 0.34]
Grade 0.25 0.06 0.25∗∗∗ [0.12, 0.39] 0.09 0.08 0.09 [−0.08, 0.26] 0.24 0.06 0.24∗∗∗ [0.11, 0.37]
Internet age 0.02 0.02 0.05 [−0.02, 0.05] 0.02 0.02 0.02 [−0.03, 0.07] 0.02 0.02 0.02 [−0.02, 0.05]
Internet time 0.03 0.02 0.03 [−0.01, 0.06] −0.03 0.02 −0.03 [−0.08, 0.02] 0.03 0.02 0.03 [−0.01, 0.07]
R2 0.54 0.20 0.56
F 33.92∗∗∗ 7.39∗∗∗ 28.53∗∗∗

Note. X, U, M, and Y represent aggressive trait, network public opinion, beliefs about aggression, and cyberbullying, respectively, as below. Therefore, the 95%
confidence interval of predicted variables was obtained by bootstrap method; ∗p-value < 0.05, ∗∗p-value < 0.01, ∗∗∗p-value < 0.001.

Fig. 3. The mediated moderator effect model. Note. R2 represents the inter-
pretation rate of variance. The dotted line is not significant; ∗p-value < 0.05, Fig. 4. The moderated mediation effect model. Note. R2 represents the inter-
∗∗
p-value < 0.01, ∗∗∗p-value < 0.001. pretation rate of variance. The dotted line is not significant; ∗p-value < 0.05,
∗∗
p-value < 0.01, ∗∗∗p-value < 0.001.

significantly negatively predicted beliefs about aggression (γ = −0.27,


considered the mediated moderator model.
p < 0.001), and beliefs about aggression significantly positively pre-
To clearly explain the nature of the effect of the interaction between
dicted cyberbullying (γ = 0.13, p < 0.05). The above results showed
aggressive traits and network public opinion on cyberbullying, a simple
that the interaction between network public opinion and aggressive
effect analysis graph was drawn based on the values of the regression
traits would indirectly affect cyberbullying through beliefs about ag-
equation taken from the mean of the independent variable and the
gression as a mediating variable. Therefore, a mediating moderator
moderating variable plus or minus one standard deviation (see Figs. 5
model is established.
and 6). It could be found that, under the guidance of network public
According to Ye and Wen (2013), if the moderated mediation model
opinion with irrational comments, aggressive traits had a significant
satisfied the following three requirements, it indicated the existence of
positive predictive effect on cyberbullying (bsimple = 0.08, SE = 0.06,
a moderated mediation effect (see Table 2). In the first step, aggressive
traits as the independent variable had a significant effect on cyberbul-
lying as the dependent variable (β = 0.47, p < 0.001). Aggressive
traits as the independent variable and network public opinion as the
moderator should significantly interact to influence cyberbullying as
the dependent variable (β = −0.24, p < 0.001). Next, as a second
step, aggressive traits as the independent variable and network public
opinion as the moderator should significantly interact to influence be-
liefs about aggression as the mediator (β = −0.28, p < 0.001). Finally,
in the third step, beliefs about aggression as the mediator and network
public opinion as the moderator should significantly interact to influ-
ence cyberbullying as the dependent variable (β = −0.12, p < 0.05),
and there should be a moderating effect of aggressive traits as the in-
dependent variable on the residual effect of cyberbullying as the de-
pendent variable (β = −0.18, p < 0.01). Predictors were standardized
in each equation, and the network age, grade, and network time were
controlled. The variance expansion factor of all predictive variables was
less than 1, so there was no multicollinearity problem. The results of
further test on the integrated model indicated that the degree of fit of
the model was not good (χ2/df = 3.02, CFI = 0.72, NFI = 0.85, Fig. 5. Interaction of aggressive traits and network public opinion on cyber-
GFI = 0.74, RMSEA = 0.078; see Fig. 4). Thus, the current study only bullying.

173
M. Song, et al. Computers in Human Behavior 97 (2019) 167–178

indirectly rather than directly online by spreading rumors and pub-


lishing false information. The older the network, the longer the time
they spend on the Internet, and the more likely they are to be bullied
(Ybarra & Mitchell, 2004).
The correlation analysis showed that aggressive traits had a sig-
nificant positive effect on cyberbullying. The higher of aggressive traits,
the more likely it might lead to cyberbullying. According to the clues
filter theory (Walther, 1992), as network communication lacks effective
social communication clues and the premise of physical absenteeism
has the characteristics of over time and space, it is impossible to cor-
rectly judge the correct tone of voice, attitude response, and emotional
state for individuals with high aggressive traits communicating with
others. This will reduce the empathy ability in these individuals without
burdening them with more negative emotions, such as guilt and re-
morse, resulting in the occurrence of cyberbullying (Allison, Bussey, &
Lindsey, 2016). In addition, when using the network, individuals with
high aggressive traits are apt to over-identify some information from
the network as threatening stimuli, which leads to cognitive imbalance
of the pre-attention system and reduces the critical value of the dan-
Fig. 6. Interaction of aggressive traits and network public opinion on beliefs
gerous stimuli probability, resulting in cognitive bias and physiological
about aggression.
alertness and further arousing anger, anxiety, and other negative
emotional experience. To adjust the imbalance caused by the rapid
p < 0.001); under the guidance of network public opinion with ra- changes of physiology and psychology, individuals often choose some
tional comments, the positive predictive effect of aggressive traits on extreme bad behaviors for self-adjustment (Barhaim, Lamy, Pergamin,
cyberbullying was weakened (bsimple = 0.03, SE = 0.09, p < 0.05). Si- Bakermanskranenburg, & van IJzendoorn, 2007), which indirectly
milarly, under the guidance of network public opinion with irrational leads to the emergence of cyberbullying.
comments, aggressive traits had a significant positive predictive effect
on beliefs about aggression (bsimple = 0.07, SE = 0.07, p < 0.001);
under the guidance of network public opinion with rational comments, 4.2. Moderating role of network public opinion
aggressive traits had no significant positive predictive effect on cyber-
bullying (bsimple = 0.01, SE = 0.08, p = 0.600)—that is, when in- After controlling gender, grade, Internet age, and Internet time, the
dividuals were in the situation of irrational expression of network current study found that Internet public opinion played a significant
public opinion, aggressive traits directly affected cyberbullying. moderating role in the relationship between aggressive traits and cy-
In summary, aggressive traits significantly positively predicted cy- berbullying. Irrational expression of Internet public opinion sig-
berbullying, network public opinion had a moderating effect on the nificantly enhanced the predictive effect of aggressive traits on cyber-
relationship between aggressive traits and cyberbullying, and beliefs bullying. Aggressive traits have relatively little predictive effect on
about aggression played a mediating role in this moderating effect. That cyberbullying in the context of the rational expression of network
is, under the guidance of the network public opinion with irrational public opinion, which supports previous findings (Dewall et al., 2011).
comments, aggressive traits not only affected cyberbullying by acting According to the trait activation theory, the effect of aggressive traits on
on beliefs about aggression but also directly affected cyberbullying. On cyberbullying will be moderated by Internet public opinion. This is
the contrary, under the guidance of the network public opinion with because the public of network public opinion (especially Internet users)
rational comments, aggressive traits directly affected the occurrence of use the Internet as a platform to exert certain influence or express
cyberbullying. biased opinions or remarks on some certain public events through
network language or other means (Hu, 2016). Because of the anonymity
4. Discussion and virtual characteristics of the current network platform, the ex-
pression of network public opinion has become differentiated into ir-
4.1. The relationship between aggressive traits and cyberbullying rational and rational expressions. Individuals with aggressive traits
influenced by the irrational expression of Internet public opinion may
The current study found no significant difference in cyberbullying as choose to maintain consistency with other people's opinions and thus be
a function of gender, which is consistent with some existing findings more likely to engage in cyberbullying, such as abuse and slander (Cao
(Zhang et al., 2017). This may be because traditional bullying relies on & Bai, 2011; Lin & Li, 2016). Näsi et al. found that there was a sig-
strength and small group implementation, while cyberbullying relies on nificant association between exposure to websites relating to eating
Internet technology and anonymity. The bullying is conducted directly disorders and online harassment. Hawdon et al. (2016) found that there
by virtue of the Internet; thus, the influence of differences in physical is a considerable amount of hate material online, but the degree to
strength between men and women is greatly reduced. In addition, the which individuals from different countries are exposed to these mate-
current study also found that grade, Internet age, and Internet time rials varies, and thus there can be country-wide differences in degree of
were significantly positively correlated with Internet bullying, which online hate. On the contrary, individuals with aggressive traits influ-
was consistent with previous findings (Li et al., 2016; Ybarra & enced by rational expressions may correspondingly show less cyber-
Mitchell, 2004). At the university stage, the goals and priorities faced bullying. Similarly, according to the “spiral of silence” theory, network
by students in each grade are also very different. These differences will public opinion is an atmosphere in which network groups comment on
lead to the stage and level of their emotions. As the grades increase, a hot topic. Under the pressure of rational or irrational comments, in-
college students who face a stronger sense of pressure, greater emo- dividuals, fearing being isolated from the group, would choose the
tional fluctuations, and more negative experiences are more inclined to viewpoint consistent with the rest of the group members by force of
vent through the Internet, which is more likely to cause cyberbullying pressure, remaining silent about their own true opinion (Zhang et al.,
(Li et al., 2016). Because of the virtual and indirect nature of the net- 2017). Especially in the irrational atmosphere of online public opinion,
work, individuals with an older age are more inclined to bully others aggressive individuals are more likely to enact cyberbullying.

174
M. Song, et al. Computers in Human Behavior 97 (2019) 167–178

4.3. Mediating role of beliefs about aggression irrational expression of network public opinion would increase beliefs
about aggression of individuals with high aggressive traits and then
The results of the current study indicated that beliefs about ag- induce cyberbullying, which has certain theoretical significance for
gression partially mediated the relationship between aggressive traits studies of aggressive traits on cyberbullying. In addition, the current
and cyberbullying. It showed that beliefs about aggression played a study proposed two competitive models and then chose a mediated
“bridge” role in the moderating effect of network public opinion on the moderator to examine the effects of aggressive traits on cyberbullying
relationship between aggressive traits and cyberbullying. That is, the as well as the moderating role of network public opinion and mediating
reason network public opinion plays a moderating role in the re- role of beliefs about aggression in this process based on research pur-
lationship between aggressive traits and cyberbullying is that the irra- pose, research focus, and integrated model fit. It not only widened the
tional expression of network public opinion plays a moderating role in application scope of general aggression model but also contributed to
the relationship between aggressive traits and beliefs about aggression. an understanding of the process of cyberbullying.
In other words, there is a mediating moderator effect between ag- In addition, the findings of the current study can provide some
gressive traits and cyberbullying; namely, beliefs about aggression play guidance for preventing and reducing cyberbullying. The current study
a mediating role in the effect of the interaction between aggressive found a positive correlation between cyberbullying and grade, network
traits and network public opinion on cyberbullying. Specifically, ag- age, and internet time. This reminds us that parents and schools should
gressive traits in the rational expression of network public opinion do correctly guide students' online behaviors, avoid indulging in network
not have much impact on beliefs about aggression and have no in- virtual space for extended periods, reasonably control online time, and
hibitory effect on cyberbullying while, in the irrational expression of reduce the potential for cyberbullying. The current study also found
network public opinion, they significantly influence cyberbullying that network public opinion moderated the relationship between ag-
through beliefs about aggression, which is consistent with the findings gressive traits and cyberbullying. This further reminds us that the
of related research (Ang et al., 2011; Williams & Guerra, 2007). Ac- government should engage with network public opinion and guide the
cording to group polarization theory, when individuals with aggressive formation of an orderly and standardized network society in cyber-
traits are in the atmosphere of irrational or rational expression of net- space. In view of the irrational comments of network public opinion, we
work public opinion, they will change their acceptance and recognition should make full use of the role of opinion leaders, change irrational
of aggressive behaviors due to the influence of emotional infection or public opinion, and create a rational atmosphere of public opinion. At
inflammatory language of other groups, thus causing cyberbullying the same time, we should improve the moral quality of netizens, purify
(Liu, 2014a, 2014b). In addition, the cognitive connection theory holds cyberspace, create a green network platform, strengthen the super-
that, when individuals with aggressive traits are in the atmosphere of vision of violent video or games on the network, and prevent copycat
irrational expression of network public opinion, aggressive words such behavior of netizens to prevent and reduce cyberbullying. Moreover,
as abuse and insult in the network platform will activate individuals’ the current study also found that beliefs about aggression mediated the
thinking memory related to aggression and affect the generation of relationship between aggressive traits and cyberbullying. At the same
cyberbullying. When network comments or opinions form an atmo- time, it suggests that we can promote a “zero tolerance” attitude to-
sphere of rational expression of network public opinion, individuals wards cyberbullying and prevent the emergence of irregular behavior
with aggressive traits will change their perceptions and judgments of by changing an individual's biased beliefs about aggressive behavior.
the current situation according to the group atmosphere, and beliefs For example, role-playing or discussion can be used to simulate the
about aggression will not be easily activated, which will reduce the adverse effects of cyberbullying on others, enhance individuals' ability
occurrence of cyberbullying accordingly (Liu et al., 2015). In the at- to empathize with the victim, and change the cognitive attitude towards
mosphere of irrational expression of network public opinion, in- aggressive behaviors, thereby reducing and preventing the occurrence
dividuals with high aggressive traits are more inclined to recognize and of cyberbullying (Zhu, Chen, & Zhou, 2014).
accept cyberbullying and consider cyberbullying as a reasonable way to However, there are still some limitations of the current study that
solve the problem. Their own cognitive bias towards rationalization of require perfection and improvement in the future. Firstly, for the se-
aggressive behaviors increases the likelihood that individuals will shift lection of independent variables, the current study only examined the
this cognitive attitude in real life to cyberspace, making them show effect of aggressive traits and beliefs about aggression in individual
more cyberbullying in the face of irrational comments or opinions in factors on cyberbullying. Future studies can examine the effect of dif-
cyberspace. ferent factors more comprehensively (such as family, school, social
The mediated moderation model can further explore the relation- environment, and so on) on cyberbullying; secondly, the measurement
ship between aggressive traits and cyberbullying, and its mechanism, of cyberbullying in the current study is only for bullies, and future
which means aggressive traits are one of the important risk factors for studies should consider other perspectives (such as bullies, bystanders)
cyberbullying, preliminarily illustrates the conditions under which ag- to explore the effects of aggressive traits on cyberbullying; thirdly, as
gressive traits work and also reveals how aggressive traits affect cy- far as the moderating mechanism is concerned, the current study only
berbullying under different conditions. The moderated mediation examined the moderator variables at the group level, and future studies
model is more effective in explaining the phenomenon of cyberbullying can explore the moderator at the individual level. Similarly, in terms of
than the pure moderation or mediation model. The current study found the mediating mechanism, the current study found that beliefs about
that, during the occurrence of cyberbullying, the irrational expression aggression only had a partial mediating effect between aggressive traits
of network public opinion in the network environment increases the and cyberbullying, which indicates that there may be other mediating
beliefs about aggression of individuals with high aggressive traits, variables. Future studies can examine the mediating variables at the
which is more likely to cause cyberbullying. On the contrary, the ra- emotional level. Fourth, the current study adopted a cross-sectional
tional expression atmosphere of network public opinion has a relatively design and questionnaire survey, which made it difficult to explain any
low probability of activating beliefs of about aggression of individuals causality between variables and could not assess the “influence” of
with aggressive traits, thus producing less cyberbullying. particular variables. Future studies may consider using a longitudinal
design to investigate the occurrence mechanism of cyberbullying from a
4.4. Practical implications and future directions developmental and dynamic perspective. Fifth, this study was based on
a small Chinese sample, and the proportion of female students was high,
Based on the general aggression model and trait activation theory, and so the effects may not be universal. Future studies should increase
the current study explored the effects of aggressive traits on cyberbul- the sample size and balance the proportion of males and females.
lying from the perspective of network public opinion and found that the

175
M. Song, et al. Computers in Human Behavior 97 (2019) 167–178

5. Conclusion 15. I am sometimes eaten up with jealousy. 1 2 3 4 5


16. I can think of no good reason for ever hitting a person.
The current study took a crucial step in exploring the roles of net- 1 2 3 4 5
work public opinion and beliefs about aggression on the impact of ag- 17. At times, I feel I have gotten a raw deal out of life. 1 2 3 4 5
gressive traits on cyberbullying and provided a substantial contribution 18. I have trouble controlling my temper. 1 2 3 4 5
to proposing measures for preventing cyberbullying among netizens. 19. When frustrated, I let my irritation show. 1 2 3 4 5
The results of the current study revealed that aggressive traits sig- 20. I sometimes feel that people are laughing at me behind my back.
nificantly predicted beliefs about aggression and cyberbullying, net- 1 2 3 4 5
work public opinion moderated the relationship between aggressive 21. I often find myself disagreeing with people. 1 2 3 4 5
traits and cyberbullying, and beliefs about aggression partially medi- 22. If somebody hits me, I hit back. 1 2 3 4 5
ated the moderating effect of network public opinion on the relation- 23. I sometimes feel like a powder keg ready to explode.
ship between aggressive traits and cyberbullying. 1 2 3 4 5
24. Other people always seem to get the breaks. 1 2 3 4 5
Author contributions 25. Given enough provocation, I may hit another person.
1 2 3 4 5
Conceive and writing frame design: LZ and SL. Wrote the paper: MS, 26. I know that “friends” talk about me behind my back.
ZZ and SL. Revise the manuscript: MS, ZZ, SL, HF, TZ and LZ. 1 2 3 4 5
27. Some of my friends think I'm a hothead. 1 2 3 4 5
Declarations of interest 28. Sometimes, I fly off the handle for no good reason. 1 2 3 4 5
29. I get into fights a little more than the average person does.
None. 1 2 3 4 5

Acknowledgments Beliefs about Aggression Scale

This research was funded by the National Natural Science Instructions: The following questions ask you about whether you
Foundation of China (71874170), the National Social Science Fund of think certain behaviors are WRONG or are OK. Please draw “○” or “√”
China (12BSH055) and the K.C. Wong Magna Fund at Ningbo that best describes what you think. Among them, 1 = not at all, 2 = a
University. Minghua Song, Zhuan Zhu and Shen Liu shared the first little, 3 = a little bit, 4 = full.
authorships. Suppose a boy says something bad to another boy, John.

Appendix A. Supplementary data 1. Do you think it's OK for John to scream at him? 1 2 3 4
2. Do you think it's OK for John to hit him? 1 2 3 4
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.chb.2019.03.015. Suppose a boy says something bad to a girl.

Appendix 3. Do you think it's wrong for the girl to scream at him? 1 2 3 4
4. Do you think it's wrong for the girl to hit him? 1 2 3 4
Aggressive Traits Scale
Suppose a girl says something bad to another girl, Mary.
Instructions: This part is the description of oneself. Please draw
“○” or “√” on the corresponding number after each sentence is read 5. Do you think it's OK for Mary to scream at her? 1 2 3 4
according to the degree of consistency with yourself. Among them, 6. Do you think it's OK for Mary to hit her? 1 2 3 4
1 = not at all, 2 = a little, 3 = not sure, 4 = a little bit, 5 = full.
Suppose a girl says something bad to a boy.
1.My friends say that I'm somewhat argumentative. 1 2 3 4 5
2.If I must resort to violence to protect my rights, I will. 7. Do you think it's wrong for the boy to scream at her? 1 2 3 4
1 2 3 4 5 8. Do you think it's wrong for the boy to hit her? 1 2 3 4
3.When people are especially nice, I wonder what they want.
1 2 3 4 5 Suppose a boy hits another boy, John?
4.I tell my friends openly when I disagree with them.
1 2 3 4 5 9. Do you think it's wrong for John to hit him back? 1 2 3 4
5. I have become so mad that I have broken things. 1 2 3 4 5
6. I can't help getting into arguments when people disagree with me. Suppose a boy hits a girl.
1 2 3 4 5
7. I wonder why sometimes I feel so bitter about things. 10. Do you think it's OK for the girl to hit him back? 1 2 3 4
1 2 3 4 5
8. Occasionally, I can't control the urge to strike another person. Suppose a girl hits another girl, Mary?
1 2 3 4 5
9. I am an even-tempered person. 1 2 3 4 5 11. Do you think it's wrong for Mary to hit her back? 1 2 3 4
10. I am suspicious of overly friendly strangers. 1 2 3 4 5
11. I have threatened people I know. 1 2 3 4 5 Suppose a girl hits a boy.
12. I flare up quickly but get over it quickly. 1 2 3 4 5
13. There are people who pushed me so far that we came to blows. 12. Do you think it's wrong for the boy to hit her back? 1 2 3 4
1 2 3 4 5 13. In general, it is wrong to hit other people. 1 2 3 4
14. When people annoy me, I may tell them what I think of them. 14. If you're angry, it is OK to say mean things to other people.
1 2 3 4 5 1 2 3 4

176
M. Song, et al. Computers in Human Behavior 97 (2019) 167–178

15. In general, it is OK to yell at others and say bad things. 1 2 3 4 intervention strategies. Aggression and Violent Behavior, 25(3), 35–42. https://doi.
16. It is usually OK to push or shove other people around if you're mad. org/10.1016/j.avb.2015.07.011.
Ang, R. P., Huan, V. S., & Florell, D. (2014). Understanding the relationship between
1 2 3 4 proactive and reactive aggression, and cyberbullying across United States and
17. It is wrong to insult other people. 1 2 3 4 Singapore adolescent samples. Journal of Interpersonal Violence, 29(2), 237–254.
18. It is wrong to take it out on others by saying mean things when https://doi.org/10.1177/0886260513505149.
Ang, R. P., Tan, K. A., & Talib, M. A. (2011). Normative beliefs about aggression as a
you're mad. 1 2 3 4 mediator of narcissistic exploitativeness and cyberbullying. Journal of Interpersonal
19. It is generally wrong to get into physical fights with others. Violence, 26(13), 2619–2634. https://doi.org/10.1177/0886260510388286.
1 2 3 4 Antoniadou, N., & Kokkinos, C. M. (2015). Cyber and school bullying: Same or different
phenomena? Aggression and Violent Behavior, 25, 363–372. https://doi.org/10.1016/
20. In general, it is OK to take your anger out on others by using j.avb.2015.09.013.
physical force. 1 2 3 4 Athanasiades, C., Baldry, A. C., Kamariotis, T., Kostouli, M., & Psalti, A. (2016). The “net”
of the internet: Risk factors for cyberbullying among secondary-school students in
Greece. European Journal on Criminal Policy and Research, 22(2), 1–17.
Cyberbullying Scale
Barhaim, Y., Lamy, D., Pergamin, L., Bakermanskranenburg, M. J., & van IJzendoorn, M.
H. (2007). Threat-related attentional bias in anxious and nonanxious individuals: A
Instructions: Many college students will participate in the inter- meta-analytic study. Psychological Bulletin, 133(1), 1–24. https://doi.org/10.1037/
action on the Internet. Weibo, and posting are very common network 0033-2909.133.1.1.
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in
behaviors. Many people will also respond to or evaluate other people's social psychological research: Conceptual, strategic, and statistical considerations.
information release. The purpose of this study is to understand college Journal of Personality and Social Psychology, 51, 1173–1182.
students' participation in network behavior and its impact on college Baroncelli, A., & Ciucci, E. (2014). Unique effects of different components of trait emo-
tional intelligence in traditional bullying and cyberbullying. Journal of Adolescence,
Students' daily life. The following are some of the behaviors on the 37(6), 807–815. https://doi.org/10.1016/j.adolescence.2014.05.009.
Internet. Please draw “○” or “√” on the corresponding number after Buss, A. H., & Perry, M. (1992). The aggression questionnaire. Journal of Personality and
each sentence is read according to the degree of consistency with Social Psychology, 63(3), 452–459. https://doi.org/10.1037/0022-3514.63.3.452.
Cao, R., & Bai, S. L. (2011). The characteristics of net opinion in China at the present
yourself. Among them, 1 = not at all, 2 = a little, 3 = not sure, 4 = a stage. Journal of Hebei University (Philosophy and Social Science), 36(2), 70–75.
little bit, 5 = full. https://doi.org/10.3969/j.issn.1000-6378.2011.02.016 [In Chinese].
Çetin, B., Yaman, E., & Peker, A. (2011). Cyber victim and bullying scale: A study of
validity and reliability. Computers & Education, 57(4), 2261–2271. https://doi.org/
1. Using offensive symbols on the Internet. 1 2 3 4 5 10.1016/j.compedu.2011.06.014.
2. Editing photos in offensive manner on the Internet. 1 2 3 4 5 Dewall, C. N., Anderson, C. A., & Bushman, B. J. (2011). The general aggression model:
3. Using Internet as a slandering tool. 1 2 3 4 5 Theoretical extensions to violence. Psychology of Violence, 1(3), 245–258. https://doi.
org/10.1037/a0023842.
4. Disclosure of information about the family members of the parties
Erceg-Hurn, D. M., & Mirosevich, V. M. (2008). Modern robust statistical methods: An
to the incident on the Internet. 1 2 3 4 5 easy way to maximize the accuracy and power of your research. American
5. Hiding identity on the Internet. 1 2 3 4 5 Psychologist, 63(7), 591–601. https://doi.org/10.1037/0003-066X.63.7.591.
6. Reproduce spoof pictures of events on the Internet. 1 2 3 4 5 Gagnon, J., Mercédès, A., Emond, F. C., Derguy, S., Brochu, A. F., Bessette, M., et al.
(2017). An ERP study on hostile attribution bias in aggressive and nonaggressive
7. Detail the event in the comments. 1 2 3 4 5 individuals: ERP study of hostile attribution bias. Aggressive Behavior, 43(3), 217–229.
8. Spread false information on the Internet. 1 2 3 4 5 https://doi.org/10.1002/ab.21676.
9. Sharing videos without permission on the Internet. 1 2 3 4 5 Geel, M. V., Goemans, A., Toprak, F., & Vedder, P. (2017). Which personality traits are
related to traditional bullying and cyberbullying? A study with the big five, dark triad
10. Sharing someone's photos without permission on the Internet. and sadism. Personality and Individual Differences, 106, 231–235. https://doi.org/10.
1 2 3 4 5 1016/j.paid.2016.10.063.
11. Writing offensive comments about news on websites. Gini, G., & Espelage, D. L. (2014). Peer victimization, cyberbullying, and suicide risk in
children and adolescents. Journal of the American Medical Association, 312(5),
1 2 3 4 5 545–546. https://doi.org/10.1001/jama.2014.3212.
12. Using humiliating expressions on the Internet. 1 2 3 4 5 Hawdon, J., Oksanen, A., & Räsänen, P. (2016). Exposure to online hate in four nations: A
13. Using Internet as a propaganda tool for own benefit. cross-national consideration. Deviant Behavior, 38(3), 254–266. https://doi.org/10.
1080/01639625.2016.1196985.
1 2 3 4 5
Hemphill, S. A., Kotevski, A., Tollit, M., Smith, R., Herrenkohl, T. I., Toumbourou, J. W.,
14. Using nicknames on the Internet in a disturbing way. et al. (2012). Longitudinal predictors of cyber and traditional bullying perpetration in
1 2 3 4 5 Australian secondary school students. Journal of Adolescent Health Official Publication
of the Society for Adolescent Medicine, 51(1), 59–65. https://doi.org/10.1016/j.
15. Publishing the personal information of the parties involved in the
jadohealth.2011.11.019.
incident through the Internet. 1 2 3 4 5 Holt, M. K., Vivolokantor, A. M., Polanin, J. R., Holland, K. M., Degue, S., Matjasko, J. L.,
16. Rumoring on the Internet. 1 2 3 4 5 et al. (2015). Bullying and suicidal ideation and behaviors: A meta-analysis.
17. Mocking on the Internet. 1 2 3 4 5 Pediatrics, 135(2), 496–509. https://doi.org/10.1542/peds.2014-1864.
Hu, Q. (2016). The correct guidance of irrational expression of network public opinion.
18. Blaming on the Internet. 1 2 3 4 5 Journal News Research, 7(11), 46.
19. Using someone's identity without his/her permission on the Huang, F., Ding, Q., Wei, H., & Hong, J. Z. (2018). Effects of post thematic characteristics
Internet. 1 2 3 4 5 on knowledge sharing in the virtual community: The bystander effect perspective.
Acta Psychology Sinica, 50(2), 226–234. https://doi.org/10.3724/SP.J.1041.2018.
20. Entering someone's private page without permission on the 00226 [In Chinese].
Internet. 1 2 3 4 5 Huesmann, L. R., & Guerra, N. G. (1997). Children's normative beliefs about aggression
21. Editing videos in offensive manner on the Internet. 1 2 3 4 5 and aggressive behavior. Journal of Personality and Social Psychology, 72(2), 408–419.
https://doi.org/10.1037/0022-3514.72.2.408.
22. Publish private life on the Internet. 1 2 3 4 5 Huesmann, L. R., & Kirwil, L. (2007). Why observing violence increases the risk of violent
23. Using abusive/insulting language in e-mails. 1 2 3 4 5 behavior by the observer. In D. J. Flannery, A. T. Vazsony, & I. Waldman (Eds.). The
24. Sharing images with sexual content on the Internet. 1 2 3 4 5 Cambridge handbook of violent behavior and aggression (pp. 545–570). Cambridge:
Cambridge University Press. https://doi.org/10.1017/CBO9780511816840.029.
25. Reprint posts on the Internet. 1 2 3 4 5
Hutson, E. (2016). Cyberbullying in adolescence: A concept analysis. Advances in Nursing
Science, 39(1), 60–70. https://doi.org/10.1097/ANS.0000000000000104.
References Jenaro, C., Flores, N., & Frías, C. P. (2017). Systematic review of empirical studies on
cyberbullying in adults: What we know and what we should investigate. Aggression
and Violent Behavior, 38, 113–122. https://doi.org/10.1016/j.avb.2017.12.003.
Adams, C., & Ireland, J. L. (2017). The role of beliefs and trait aggression in prison Jiang, W. S., & Liu, X. R. (2011). Analysis of irrational expression in the formation of
bullying among young offenders. Journal of Forensic Psychiatry and Psychology, 29(3), network public opinion. Southeast Communication, 8(6), 41–43. https://doi.org/10.
1–17. https://doi.org/10.1080/14789949.2017.1395060. 3969/j.issn.1672-9579.2011.06.014 [In Chinese].
Allison, K. R., Bussey, K., & Lindsey, D. (2016). Cyber-bystanding in context: A review of Keipi, T., Näsi, M., Oksanen, A., & Räsänen, P. (2016). Online hate and harmful content:
the literature on witnesses' responses to cyberbullying. Children and Youth Services Cross-national perspectives. Routledgehttps://doi.org/10.4324/9781315628370.
Review, 65, 183–194. https://doi.org/10.1016/j.childyouth.2016.03.026. Kowalski, R. M., Giumetti, G. W., Schroeder, A. N., & Lattanner, M. R. (2014). Bullying in
Ang, R. P. (2015). Adolescent cyberbullying: A review of characteristics, prevention and the digital age: A critical review and meta-analysis of cyberbullying research among

177
M. Song, et al. Computers in Human Behavior 97 (2019) 167–178

youth. Psychological Bulletin, 140(4), 1073–1137. https://doi.org/10.1037/ model of cyberbullying perpetration, aggression, social skills, and internet self-effi-
a0036634. cacy. Computers in Human Behavior, 71, 353–361. https://doi.org/10.1016/j.chb.
Li, Q. (2006). Cyberbullying in schools: A research of gender differences. School 2017.02.016.
Psychology International, 27(2), 157–170. https://doi.org/10.1177/ Selkie, E. M., Kota, R., Chan, Y. F., & Moreno, M. (2015). Cyberbullying, depression, and
0143034306064547. problem alcohol use in female college students: A multisite study. Cyberpsychology,
Li, W., Li, K., & Zhang, Q. C. (2016). Narcissism and cyber-bullying: A mediating effect of Behavior, and Social Networking, 18(2), 79–86. https://doi.org/10.1089/cyber.2014.
moral disengagement. Psychology: Techniques and Application, 4(11), 676–683. 0371.
https://doi.org/10.16842/j.cnki.issn2095-5588.2016.11.006 [In Chinese]. Slonje, R., & Smith, P. K. (2008). Cyberbullying: Another main type of bullying?
Lin, J. J., & Li, Z. C. (2016). The reasons and avoidance of irrational network public Scandinavian Journal of Psychology, 49(2), 147–154. https://doi.org/10.1111/j.1467-
opinion. Young Journalist, 76(2), 26–27 [In Chinese] doi: 103969/j.issn.1002- 9450.2007.00611.x.
2759.2016.02.017. Smith, P., Waterman, M., & Ward, N. (2011). Driving aggression in forensic and non-
Liu, H. Z. (2014a). On the irrational network public opinion on group events and its forensic populations: Relationships to self-reported levels of aggression, anger and
psychological mechanism. Journal of Poltical Science and Law, 31(6), 81–85 [In impulsivity. British Journal of Psychology, 97(3), 387–403. https://doi.org/10.1348/
Chinese]. 000712605X79111.
Liu, J. (2014b). Peeking into the psychology of netizens and guiding network public Tett, R. P., & Burnett, D. D. (2003). A personality trait-based interactionist model of job
opinion: A study of the guidance of network public opinion in the perspective of performance. Journal of Applied Psychology, 88(3), 500–517. https://doi.org/10.
psychology. News of the World, 26(11), 150–151 [In Chinese]. 1037/0021-9010.88.3.500.
Liu, W. W., Jiang, Q., Ren, J. J., Li, S. F., & Xu, Y. P. (2015). The impact of trait anger on Topcu, C., & Erdurbaker, O. (2012). Affective and cognitive empathy as mediators of
aggressive behavior: Moderated mediating effect of hostile cognition and impulsivity gender differences in cyber and traditional bullying. School Psychology International,
level. Psychological Development and Education, 31(4), 485–493. https://doi.org/10. 33(5), 550–561. https://doi.org/10.1177/0143034312446882.
16187/j.cnki.issn1001-4918.2015.04.13 [In Chinese]. Tremblay, P. F., & Belchevski, M. (2004). Did the instigator intend to provoke? A key
Long, X. D. (2014). The impact of network group psychology on network public opinion. moderator in the relation between trait aggression and aggressive behavior.
Today’ Massmedia, 23(7), 31–32. https://doi.org/10.3969/j.issn.1672-8122.2014.07. Aggressive Behavior, 30(5), 409–424. https://doi.org/10.1002/ab.20027.
014 [In Chinese]. Walther, J. B. (1992). Interpersonal effects in computer-mediated interaction: A relational
Maier, K. J., & James, A. E. (2014). Hostility and social support explain physical activity perspective. Communication Research, 19(1), 52–90. https://doi.org/10.1177/
beyond negative affect among young men, but not women, in college. Behavioral 009365092019001003.
Medicine, 40, 34–41. https://doi.org/10.1080/08964289.2013.826170. Wen, Z., Chang, L., & Hau, K. T. (2006). Mediated moderator and moderated mediator.
Molapour, T., Lindström, B., & Olsson, A. (2016). Aversive learning and trait aggression Acta Psychologica Sinica, 38(3), 448–452.
influence retaliatory behavior. Frontiers in Psychology, 7, 833–843. https://doi.org/ Wen, Z. L., Liu, H. Y., & Hau, K. T. (2012). Analyses of moderating and mediating effects.
10.3389/fpsyg.2016.00833. Beijing: Educational Science Publishing House.
Montuoro, P., & Mainhard, T. (2017). An investigation of the mechanism underlying Wilkowski, B. M., & Robinson, M. D. (2008). The cognitive basis of trait anger and re-
teacher aggression: Testing I(3) theory and the general aggression model. British active aggression: An integrative analysis. Personality and Social Psychology Review,
Journal of Educational Psychology, 87(4), 497–517. https://doi.org/10.1111/bjep. 12(1), 3–21. https://doi.org/10.1177/1088868307309874.
12161. Williams, K. R., & Guerra, N. G. (2007). Prevalence and predictors of internet bullying.
Muller, D., Judd, C. M., & Yzerbyt, V. Y. (2005). When moderation is mediated and Journal of Adolescent Health, 41(6), S14–S21. https://doi.org/10.1016/j.jadohealth.
mediation is moderated. Journal of Personality and Social Psychology, 89(6), 852–863. 2007.08.018.
https://doi.org/10.1037/0022-3514.89.6.852. Wright, M. F., & Li, Y. (2013). Normative beliefs about aggression and cyber aggression
Paananen, M., Aro, T., Viholainen, H., Koponen, T., Tolvanen, A., Westerholm, J., & Aro, among young adults: A longitudinal investigation. Aggressive Behavior, 39(3),
M. (2019). Self-regulatory efficacy and sources of efficacy in elementary school pu- 161–170. https://doi.org/10.1002/ab.21470.
pils: Self-regulatory experiences in a population sample and pupils with attention and Xing, X. D. (2015). The social psychological origin of irrational network public opinion.
executive function difficulties. Learning and Individual Differences, 70, 53–61. https:// Youth Journalist, 75(35), 10–11. https://doi.org/10.3969/j.issn.1002-2759.2015.35.
doi.org/10.1016/j.lindif.2019.01.003. 006 [In Chinese].
Palermiti, A. L., Servidio, R., Bartolo, M. G., & Costabile, A. (2017). Cyberbullying and Ybarra, M. L., & Mitchell, K. J. (2004). Online aggressor/targets, aggressors, and targets:
self-esteem: An Italian study. Computers in Human Behavior, 69, 136–141. https://doi. A comparison of associated youth characteristics. Journal of Child Psychology and
org/10.1016/j.chb.2016.12.026. Psychiatry, 45(7), 1308–1316. https://doi.org/10.1111/j.1469-7610.2004.00328.x.
Peter, I. K., & Petermann, F. (2018). Cyberbullying: A concept analysis of defining at- Ye, B. J., & Wen, Z. L. (2013). A discussion on testing methods for mediated moderation
tributes and additional influencing factors. Computers in Human Behavior, 86, models: Discrimination and integration. Acta Psychology Sinica, 45(9), 1050–1060.
350–366. https://doi.org/10.1016/j.chb.2018.05.013. https://doi.org/10.3724/SP.J.1041.2013.01050 [In Chinese].
Podsakoff, P. M., Mackenzie, S. B., Lee, J. Y., & Podsakoff, N. P. (2003). Common method Zhang, L., Liu, S., Xu, Q., Wu, X. Y., & Yang, M. Y. (2017). Long-term effect of violence
biases in behavioral research: A critical review of the literature and recommended exposure in real-life on aggressive behaviors: A moderated mediation model. Acta
remedies. Journal of Applied Psychology, 88(5), 879–903. https://doi.org/10.1037/ Psychology Sinica, 49(1), 50–59. https://doi.org/10.3724/SP.J.1041.2017.00050 [In
0021-9010.88.5.879. Chinese].
Poon, K. T., & Chen, Z. S. (2014). When justice surrenders: The effect of just-world beliefs Zhang, Z. A., & Yan, Q. H. (2016). Individual emotion, social emotion and collective will:
on aggression following ostracism. Journal of Experimental Social Psychology, 52, On the irrationality and its factors of network public opinion. Journalism Review,
101–112. https://doi.org/10.1016/j.jesp.2014.01.006. 34(11), 16–22. https://doi.org/10.16057/j.cnki.31-1171/g2.2016.11.008 [In
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing Chinese].
and comparing indirect effects in multiple mediator models. Behavior Research Zheng, Q., Ye, B. J., Yao, Y. M., Chen, J. W., Fu, H. H., Lei, X., et al. (2017). Normative
Methods, 40(3), 879–891. https://doi.org/10.3758/BRM.40.3.879. beliefs about aggression on cyberbullying: The mediating roles of moral disengage-
Reed, K. P., Cooper, R. L., Nugent, W. R., & Russell, K. (2016). Cyberbullying: A literature ment and internet morality. Chinese Journal of Clinical Psychology, 25(4), 727–730.
review of its relationship to adolescent depression and current intervention strate- https://doi.org/10.16128/j.cnki.1005-3611.2017.04.029 [In Chinese].
gies. Journal of Human Behavior in the Social Environment, 26(1), 37–45. https://doi. Zhen, S., Xie, H., Zhang, W., Wang, S., & Li, D. (2011). Exposure to violent computer
org/10.1080/10911359.2015.1059165. games and Chinese adolescents' physical aggression: The role of beliefs about ag-
Rey, R. D., Lazuras, L., Casas, J. A., Barkoukis, V., Ortega-Ruiz, R., & Tsorbatzoudis, H. gression, hostile expectations, and empathy. Computers in Human Behavior, 27(5),
(2016). Does empathy predict (cyber) bullying perpetration, and how do age, gender 1675–1687. https://doi.org/10.1016/j.chb.2011.02.006.
and nationality affect this relationship? Learning and Individual Differences, 45, Zhu, Y. H., Chen, Q., & Zhou, H. Z. (2014). Review and latest progress of research on
275–281. https://doi.org/10.1016/j.lindif.2015.11.021. cyberbullying abroad. Chinese Youth Studies, 26(11), 80–85. https://doi.org/10.
Savage, M. W., & Tokunaga, R. S. (2017). Moving toward a theory: Testing an integrated 3969/j.issn.1002-9931.2014.11.016 [In Chinese].

178

View publication stats

You might also like