Computers in Human Behavior 110 (2020) 106338

Contents lists available at ScienceDirect

Computers in Human Behavior

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

Full length article

How students react to different cyberbullying events: Past experience,

judgment, perceived seriousness, helping behavior and the effect of
online disinhibition
Chiao Ling Huang a, Sining Zhang a, Shu Ching Yang b, *
a
Faculty of Education, Department of Educational Information Technology, East China Normal University, Shanghai, China
b
Graduate Institute of Education, National Sun Yat-sen University, Kaohsiung, Taiwan

A R T I C L E I N F O A B S T R A C T

Keywords: This study designed three scenarios related to cyberbullying (CB) to examine the relationship between self-
Cyberbullying reported CB experience, overall judgment of CB, perception of CB seriousness, helping behavior, and two
Online disinhibition types of online disinhibition (benign and toxic). A total of 415 Chinese students in three different educational
Helping behavior
stages (junior high school, senior high school and university) completed a questionnaire. The results showed that
Judgment
compared to girls, boys were more likely to bully others on the Internet, and they perceived the toxic disinhi­
bition effect more strongly. In contrast, girls more easily recognized the behaviors that occurred in harassment
and denigration scenarios as bullying instances, and they tended to have strong judgments of these behaviors.
Although no gender difference was observed in helping behavior, junior high school students had higher in­
tentions to protect victims than university students in denigration and exclusion scenarios. Correlation analysis
revealed that almost all variables were closely connected with each other, and the regression models of our
research variables effectively predicted three forms of CB. Perception of seriousness and toxic disinhibition were
common predictors in these scenarios. Based on the results, suggestions for reducing bullying incidents and
future research directions are provided.

Funding they used to be a cyber-victim. This issue has been recognized as a global
public health challenge (Huang, Yang, & Hsieh, 2019), because the
This study was sponsored by Fundamental Research Funds for the features of cyberworld such as the malignancy, continuity, perceived
Central Universities, China (Grant number 2018ECNU-HLYT019). invincibility and anonymity make it more penetrable and lethal than
bullying in reality (Willard, 2007). These experiences have a profound
1. Introduction impact on victims, influencing their psychology, behavior, health, and
suicidal intentions (Brochado, Soares, & Fraga, 2017).
The generation of digital-native students has been exposed to Evidence shows that cyberbullying is not specific to an educational
Internet and technology products since childhood. Members of this stage; it can occur in age groups ranging from primary school students to
generation are accustomed to connecting with others through digital college students. Unfortunately, most of the research on cyberbullying
devices such as computers and mobile phones. They use communication has focused on North America or Europe (Brochado et al., 2017; Chan &
software and social networking sites to build their circle of friends. Their Wong, 2015; Leung, Wong, & Farver, 2018). Relatively few studies have
social lives are inseparably related to information and communication examined this phenomenon among Chinese students (Chan & Wong,
technology (ICT). Therefore, the risks associated with social networking 2015; Zhou et al., 2013), leaving a gap in this field. In addition, because
are receiving increasing attention. this topic is sensitive, students may be deeply affected by social expec­
Among many risks, the prevalence of cyberbullying may over our tations. They are likely to underreport their actual bullying experiences
estimation. For example, a recent 2019 study conducted by Al-Rahmi or may be reluctant to disclose their attitudes toward and perceptions of
et al. indicated that more than half of their respondents admitted that bullying. Thus, future research needs to inspect the cyberbullying

* Corresponding author. Graduate Institute of Education, National Sun Yat-sen University, 70 Lien-hai Rd. Kaohsiung, R.O.C. 80424, Taiwan.
E-mail address: [email protected] (S.C. Yang).

https://doi.org/10.1016/j.chb.2020.106338
Received 3 December 2019; Received in revised form 9 February 2020; Accepted 8 March 2020
Available online 9 March 2020
0747-5632/© 2020 Elsevier Ltd. All rights reserved.
C.L. Huang et al. Computers in Human Behavior 110 (2020) 106338

involvement under Chinese context and determine how to avoid the 2. Literature review
social desirability effect on students’ answers.
The characteristics of bullies and victims have received much 2.1. The status quo of cyberbullying in China and the demographic
attention. Empirical studies have indicated that gender (Barlett & characteristics of bullies and victims
Coyne, 2014; Kowalski & Limber, 2007; Lapidot-Lefler & Dolev-Cohen,
2015; Resett & Gamez-Guadix, 2017), age (Barlett & Coyne, 2014; With the popularity of computers and networks in China, the risk
Kowalski & Limber, 2007) and other individual traits affect bullying behaviors generated by Internet use have begun to attract the attention
behaviors. In addition to the two most easily observed roles of the bully of Chinese researchers. Researchers have increasingly explored the
and the victim, researchers have noted the importance of the role of phenomenon of Chinese students’ cyberbullying. In recent studies, Li,
bystanders during a bullying incident. In a study by Salmivalli, Lager­ Luo, Lin, and Shadiev (2018) examined the relationship between
spetz, Bjorkqvist, Osterman and Kaukialnen (1996), bystanders were Internet use activities and bullying experience with 2327 Chinese stu­
distinguished as reinforcers of the bully, assistants to the bully, de­ dents (aged 9–22 years) and found that 13.9% of students were involved
fenders of the victim, and outsiders. A protector refers to a person who in at least one cyberbullying incident, with 6.2% of them acting as
stops bullying behavior and comforts the victim (Liu & Tung, 2018), and bullies, and of these bullies, 69.9% were boys and 30.1% were girls.
a protector is a vital factor in preventing bullying (Song & Oh, 2018). Leung et al. (2018) analyzed potential cyberbullying-related vari­
Individuals can intervene to stop or mitigate a cyberbullying incident by ables with 312 Hong Kong university students (aged 18–25 years) and
providing support (e.g., Brody & Vangelisti, 2017; DeSmet et al., 2016). found that 58% of the participants admitted to cyberbullying others, and
Therefore, it is imperative to adopt active means to decrease the of those perpetrators, 68% also reported being cybervictimized. Yang,
occurrence of cyberbullying via educational interventions designed to Wang, Chen, and Liu (2018) examined the cyberbullying experience of
turn students from spectators into defenders (Song & Oh, 2018). Given, 649 Chinese teenagers (aged 11–19 years), analyzing the effects of moral
to design a proper intervention seems to be a direct and effect way to separation, moral identity and parental conflict on cyberbullying
stop cyberbullying, we included gender and education stage as research behavior. They found that 17.6% of teenagers had cyberbullied others
variables in this study because it is relatively feasible for education during the past seven days before the investigation. The above surveys
practitioners to design specified intervention programs based on these showed that despite the different percentages found in the data, Chinese
two traits. students do face the risk of cyberbullying when using the Internet.
It is worth noting that the helping behavior of bystanders is closely Regarding the demographic characteristics of bullies and victims,
connected to the perceived seriousness of cyberbullying. A highly severe gender and age are often considered factors in cyberbullying research.
cyberbullying event triggers witnesses’ sense of responsibility and in­ As mentioned before, Li et al. (2018) found that boys were more likely to
fluences their willingness to help victims (Obermaier, Fawzi, & Koch, be perpetrator online. Other research also indicated that gender differ­
2016). In other words, the perceived severity of cyberbullying is a ences existed in bullying activities. For example, Kowalski and Limber
critical situational element that influences whether a bystander takes (2007) surveyed 3767 primary and middle school students in grades 6–8
action (Song & Oh, 2018) and should be included in these discussions. in the United States and found gender differences in the frequency of
In addition to the above variables, the online disinhibition effect may electronic bullying, with girls outnumbering boys in terms of bullying
deserve our attention. Joinson (1998) defines “disinhibition as lack of others and being both a bully and a victim. In addition, compared to 7th
inhibition or a type of behavior that is not constrained or restrained, and 8th grade students, 6th graders were less likely to be cyberbullying
implying a reduction in concerns for self-representation and the judg­ perpetrators or victims. Lapidot-Lefler and Dolev-Cohen (2015)
ment of others (as cited in Udris, 2014, p254).” Online disinhibition can compared the cyberbullying and traditional bullying experiences of 465
have positive or negative effects on individuals and can be classified into 7th to 12th graders in Israel and found that boys tended to bully others
benign or toxic disinhibition according to the nature of the behavior. more than girls online. However, there were no gender differences be­
Since online disinhibition leads individuals to behave differently in tween online victims and bystanders and no significant difference in the
real-world and online domains, researchers have started to study the cyberbullying behavior between middle school students and high school
relationship between it and cyberbullying. They have found that benign students. Barlett and Coyne (2014) conducted a meta-analysis of 109
disinhibition (Udris, 2014) and toxicity disinhibition (Charaschanya & studies and found that boys were more likely to be cyberbullies than
Blauw, 2017; Udris, 2014) are closely connected to cyberbullying. girls, but this trend was moderated by age. Specifically, girls were more
Nonetheless, the literature on this topic is still scant, providing a ratio­ likely to participate in cyberbullying activities in early and middle
nale for follow-up study. adolescence, and boys were more likely to do so in late adolescence. In
Although researchers have devoted themselves to studying this field view of these studies, we find that demographic characteristics play an
and have obtained fruitful outcomes, the existing research has examined important role in individuals’ bullying or victimization experiences,
the relationship between individual characteristics, helping behavior, even if the directions of influence vary.
perception of the seriousness of cyberbullying, online disinhibition and
cyberbullying experience separately and have rarely considered the 2.2. Helping behavior, perceived seriousness, and cyberbullying
possible multivariate relationships among these variables. However,
individual behavior is a decision that is intertwined with many factors Many factors lead bystanders to intervene in cyberbullying. One
that should be considered together when discussing a specific behavior. factor is the perceived severity of bullying events. The higher the
In particular, the literature has demonstrated that some of these vari­ perceived seriousness, the more willing bystanders are to intervene (e.g.,
ables are closely connected. In these circumstances, studying these Bastiaensens et al., 2014; Koehler & Weber, 2018; Obermaier et al.,
variables in isolation will provide fragmented results which impede a 2016).
full comprehension of cyberbullying, and furthermore, make it difficult For example, Bastiaensens et al. (2014) investigated the potential
to design proper interventions or provide expedient assistance to stu­ situational factors that influence students to become protectors or fa­
dents. In addition, developing a new instrument that can effectively cilitators of cyberbullying activities. They conducted 2 (low vs. high
evaluate students’ actual attitudes and experience and inspect the perceived severity) � 2 (other bystanders reinforce the bully vs. defend
bullying situation in an Eastern context is imperative. To bridge these the victim) � 2 (other bystanders are acquaintances vs. good friends)
research gaps, this study developed three bullying scenarios to explore factorial experiment. The results revealed that bystanders were more
the relationships among the above variables in Chinese students, and it likely to help victims who were severely harassed than victims who were
also examined their judgment of cyberbullying to enrich our under­ more mildly harassed. Koehler and Weber (2018) recruited 240 German
standing of the relevant factors. participants with an average age of 21.3 years to clarify how the

2
C.L. Huang et al. Computers in Human Behavior 110 (2020) 106338

different severity levels of cyberbullying events affected bystanders’ context (Suler, 2004).
perceptions, tendency to blame victims, and response to the victimiza­ Evidence suggests that online disinhibition is associated with
tion. The results indicated that the more serious the incident was, the cyberbullying. For example, Udris (2014) created the Online Disinhi­
more willing bystanders were to help the victim. In addition, bystanders bition Scale (ODS) and examined the relationship between online
were inclined to blame victims in a less serious context. Obermaier et al. disinhibition and cyberbullying. After analyzing data on 887 Japanese
(2016) analyzed the reactions of 440 participants in scenarios of varying middle school students aged 16.31 years on average, they found that
cyberbullying severity levels (medium and high) and found that the benign and toxic disinhibition can significantly predict a student’s past
perception of the seriousness of cyberbullying indirectly predicted the cyberbullying experience. Charaschanya and Blauw (2017) used
intention to intervene. The authors explained that the severity of cyberbullying experience as a mediator to explore the relationship be­
cyberbullying made students recognize the urgency of the situation and tween online disinhibition and negative psychological characteristics.
enhanced their sense of responsibility, which would subsequently lead After surveying 217 Thai university students aged 17–42 years, they
them to help. Therefore, when discussing helping behaviors in cyber­ found that toxic disinhibition could directly predict cyberbullying,
bullying events, the severity perceived by bystanders should be depression and stress with the experience of victimization and that the
considered in order to avoid fragmented or incomplete accounts of the experience of bullying and victimization could predict depression and
interaction. perceived pressure. Wright, Harper, and Wachs (2019) scrutinized the
In addition, evidence has suggested that bystanders’ behaviors are moderating role of online disinhibition in the associations among ado­
related to their past victimization experiences, but the direction of the lescents’ callous-unemotional traits (callousness, uncaring, unemo­
relationship has not yet been finalized. For example, Cleemput, Van­ tional) and anonymous and non-anonymous cyberbullying. They
debosch, and Pabian (2014) inspected several determinants of bystander surveyed 1047 American 7th and 8th graders and found an interaction
behavior and their links with personal and contextual factors. They among uncaring traits and anonymous and non-anonymous cyberbul­
analyzed 2333 self-reports from Flemish 9 to 16 year-olds and found that lying under the effect of online disinhibition.
respondents who helped the victim were younger, had higher levels of Notably, anonymity is a critical element of online disinhibition that
empathy and were more likely to have been a victim of cyberbullying or supports the relationship between online disinhibition and bystander
traditional bullying in recent months. Adolescents who did nothing behavior. “The anonymity can provide added confidence for digital
when they witnessed cyberbullying were older, showed lower levels of bystanders to actively report the cyberbullying act without the threat of
empathy and were less likely to have been a victim of traditional retaliation and/or can also serve as justification to remain a passive
bullying than those who intervened. Cao and Lin (2015) examined 622 bystander without the fear of judgment by a known or unknown cyber-
teenage (aged 12–17) SNS users’ responses to the Pew Internet Survey audience” (Wong-Lo & Bullock, 2014, p. 420). Therefore, the role of
regarding their experiences of victimization and their reaction after online disinhibition should not be overlooked in the examination of
watching a bullying incident. They found that those who had been either cyberbully or bystander behavior.
victims of cyberbullying reported more antisocial reactions (e.g., joining
the bullying activity) than non-victims. Interestingly, a recent study 3. Method
found that individuals’ reaction as bystanders relates to their own
bullying behavior may show cultural differences. Ferreira, Simao, Fer­ 3.1. Sample
reira, Souza, and Francisco (2016) compared the bystander behaviors of
788 Portuguese and Brazilian college students and found that Brazilian This study adopted convenience sampling and conducted question­
bystanders who were inactive were more also likely to be victims or naires in the form of network testing. Snowball sampling was used with
aggressors themselves, whereas those who intervened were less likely to friends, and students were recruited through online social groups.
be victims or aggressors. Recruitment links were provided under the recruitment instructions to
Positive bystander behavior has been confirmed by many scholars as facilitate interested parties to directly open the webpage to provide re­
a good protective factor for cyberbullying as it can effectively reduce sponses. The webpage explained the purpose of the study, the way it
bullying damage by stopping bullies or helping victims (e.g., Brody & would be conducted, and the time for responding. The content clearly
Vangelisti, 2017; DeSmet et al., 2016). Unfortunately, bystanders are conveyed that the study was voluntary, anonymous, and confidential,
less willing to intervene in cyberbullying than in traditional bullying that all information was for academic analysis only and that the results
(Barlinska, Szuster, & Winiewski, 2013; Quirk & Campbell, 2015; would be presented in a holistic manner. Subjects could choose to
Schacter, Greenberg, & Juvonen, 2016). Additionally, the existing participate or not and could abandon the survey at any time.
literature focuses on examining the relationship between bystander Of 499 questionnaires collected, 415 were retained for analysis after
behavior and victim experience and puts less emphasis on the associa­ incomplete questionnaires were discarded; thus, the effective sample
tion between bystanders’ reaction and their own cyberbullying, leaving rate was 83.17%. Regarding the characteristics of the respondents, 142
a gap in this field. Since many studies have revealed that students may were males and 273 were females. They were students in middle school
be the bully and the victim at the same time, it is meaningful to examine (n ¼ 101), in high school (n ¼ 58), or at a university (n ¼ 256). In
the relationships between past bullying experience and bystander addition, 38.80% (n ¼ 161) of respondents spent an average of more
behaviors. than 7 h online weekly, 17.59% (n ¼ 73) 5–7 h, 17.11% (n ¼ 71) 3–5 h,
19.28% (n ¼ 80) 1–3 h, and 7.23% (n ¼ 30) less than 1 h.
2.3. Online disinhibition and cyberbullying
3.2. Instrument
Disinhibition can work in two seemingly opposing directions, as it
can be benign or toxic (Suler, 2004; Willard, 2003). The former refers to 3.2.1. Online disinhibition scale (ODS)
the promotion of openness, kindness and generosity; the latter involves We adopted the ODS developed by Udris (2014) to evaluate the
negative behaviors such as rude language, expressions of hatred, threats extent of online disinhibition perceived by students. The ODS has 11
or exploring biased information that has not been revealed in the offline items with a four-option response format, ranging from 0 (disagree) to 3

3
C.L. Huang et al. Computers in Human Behavior 110 (2020) 106338

Table 1 Table 2
Descriptions of the three cyberbullying scenarios. Gender difference across three scenarios.
Scenario A: Harassment Mean SD t p
Sonia and Debbie often hang out together. Recently, Sonia asked Debbie to do her a
Male Female Male Female
favor, but Debbie refused. Sonia thought Debbie was not a good friend. Therefore,
she repeatedly sent messages to insult her as an emotional catharsis. Scenario A: Harassment
Claim: Sonia believed that Debbie had to pay for her behavior. Cyberbullying 1.96 1.47 1.20 1.00 4.12 <.001
Scenario B: Denigration Helping behavior 5.65 5.65 1.13 1.21 .05 ¼ .957
Ed found out that Tina likes Flynn, and he shouted this secret from the rooftop. Tina Severity 5.62 6.15 1.37 1.23 3.86 <.001
got angry and struck back at him by spreading a fake scandal about Ed online. Many Judgement 5.96 6.36 1.20 1.17 3.29 ¼ .001
netizens who did not know the truth also joined in the discussion. Scenario B: Denigration
Claim: Tina believed that information on the Internet is different from information Cyberbullying 1.98 1.44 1.26 .99 4.40 <.001
in reality and that this information would not have any impact on Ed. Helping behavior 5.59 5.55 1.26 1.35 .35 ¼ .726
Scenario C: Exclusion Severity 5.85 6.18 1.28 1.26 2.53 ¼ .012
Max is a class leader. He and his classmates use an instant messaging app to Judgement 5.82 6.32 1.35 1.14 3.98 <.001
communicate about school matters. Recently, Ada made a mistake and missed an Scenario C: Exclusion
important group project. To prevent new problems from cropping up unexpectedly, Cyberbullying 1.82 1.52 1.27 1.02 2.50 ¼ .013
Max asked other classmates to keep her out of all discussions related to school Helping behavior 5.75 5.81 1.24 1.23 -.46 ¼ .648
affairs. Severity 5.80 6.25 1.30 1.18 3.52 <.001
Claim: Max believed that Ada would involve others in problems and that this was Judgement 5.96 6.22 1.37 1.33 1.87 ¼ .063
the best way to avoid more problems. Benign disinhibition 2.97 2.87 .54 .55 1.80 ¼ .073
Toxic disinhibition 1.90 1.41 .98 .67 5.35 <.001

(agree). The scale has two dimensions: benign disinhibition (e.g., the Levene’s or Mauchly’s test), we adopted Welch’s test and Pillai’s trace to
Internet is anonymous so it is easier for me to express my true feelings or compare the mean values. Finally, Pearson correlation analysis and
thoughts) and toxic disinhibition (e.g., It is easy to write insulting things multiple regression were applied to inspect the relationship between all
online because there are no repercussions). According to Udris’ (2014) variables used in this study.
report, the CFA fit indices are TLI ¼ .99, CFI ¼ .99, and RMSEA ¼ .05;
the alpha values of benign disinhibition and toxic disinhibition are .81 4. Findings
and .85, respectively. The alpha values obtained in our study sample
were .74 and .92. 4.1. The differences in gender and educational stages between
cyberbullying experience, helping behavior, perceived seriousness,
3.2.2. Three cyberbullying scenarios judgment and online disinhibition
We designed three cyberbullying scenarios (harassment, denigration
and exclusion) to assess students’ perception of seriousness, judgment, Table 2 shows the average mean scores and SD for all research var­
helping behavior and bullying experience (see Table 1). Because the iables. As observed, the scores in the three types of cyberbullying ex­
hypothetical nature of the scenario method can minimize the sense of periences were quite low.
judgment and the social expectation (Fowler, 1988), this method was Boys were generally more likely than girls to bully others in the form
used to decrease the social desirability effect. of harassment (t ¼ 4.12, p < .001), denigration (t ¼ 4.40, p < .001) and
A seven-point Likert-type scale with response options ranging from 1 exclusion (t ¼ 2.50, p ¼ .013) and were more deeply affected by toxic
(strongly disagree), 2 (disagree), 3 (somewhat disagree), 4 (neutral), 5 disinhibition (t ¼ 5.35, p < .001). Girls were more likely to view the
(somewhat agree), 6 (agree) to 7 (strongly agree) was used in this study. incidents described in the harassment (t ¼ 3.29, p ¼ .001) and deni­
Each scenario comprised six items: one item for seriousness (e.g., This is gration scenes (t ¼ 3.98, p < .001) as bullying events, and they
a serious matter), three items for helping behavior (e.g., I will report this perceived the scenarios to be more serious than boys. Girls were also
incident to a teacher), one item for overall judgment (e.g., This is more likely to believe that exclusion was unacceptable (t ¼ 3.52, p <
cyberbullying) and one item for cyberbully experience (e.g., I have al­ .001). However, no gender difference existed in helping behavior be­
ways done what the protagonist did). It is worth noting that, with the tween these three scenarios.
exception of helping behavior, the remaining factors were only assessed Regarding educational stages, the results showed that students
with one question in these scenarios. Therefore, we could not calculate differed in their attitudes toward the severity of harassment (Welch
the internal consistency coefficient for single scenario. Instead, we value ¼ 3.24, p ¼ .042) as well as in their tendencies to help victims in
assessed the overall reliability by calculating the sum of the three sce­ the denigration (F ¼ 5.33, p ¼ .005) and exclusion (Welch value ¼ 3.23,
narios with the same dimension, and the alpha values were .82 (seri­ p ¼ .043) scenarios. Specifically, junior high school students evaluated
ousness), .87 (helping behavior), .77 (overall judgment) and .87 harassment as more serious and were also more willing than university
(cyberbully experience). students to help victims of denigration and exclusion (see Table 3).
Notably, if we consider the three scenarios simultaneously, we find
3.2.3. Data analysis that students were more likely to help victims of exclusion than victims
We performed independent t-test, one-way ANOVA, repeated mea­ of harassment and denigration (Mauchly’s w ¼ .97, p ¼ .002; Pillai’s
sures ANOVA, correlation analysis and multiple regression. The t-test trace ¼ .04, F ¼ 9.40, p < .001).
and one-way ANOVA were used to analyze gender and educational stage
differences in all variables in each scenario. Repeated measures ANOVA
was conducted to test the differences in all variables across the three
scenarios. For data violating the homogeneity assumption (according to

4
C.L. Huang et al. Computers in Human Behavior 110 (2020) 106338

Table 3
Educational stage difference across three scenarios.
Mean SD Welch p

Junior Senior University Junior Senior University

Scenario A: Harassment
Cyberbullying 1.57 1.84 1.62 1.05 1.50 1.01 .75 ¼ .476
Helping behavior 5.85 5.73 5.55 1.24 1.45 1.08 2.30 ¼ .104
Severity 6.24 5.91 5.87 1.23 1.68 1.21 3.24 ¼ .042
Judgement 6.43 6.05 6.18 1.06 1.63 1.12 2.29 ¼ .105
Scenario B: Denigration
Cyberbullying 1.53 1.97 1.59 1.14 1.73 .91 1.48 ¼ .231
Helping behavior 5.87 5.74 5.40 1.23 1.54 1.28 F (5.33) ¼ .005
Severity 6.19 6.03 6.02 1.43 1.57 1.13 .57 ¼ .569
Judgement 6.32 6.02 6.11 1.29 1.62 1.12 1.20 ¼ .303
Scenario C: Exclusion
Cyberbullying 1.66 2.03 1.51 1.22 1.67 .88 3.09 ¼ .050
Helping behavior 6.06 5.80 5.68 1.34 1.61 1.07 3.23 ¼ .043
Severity 6.13 6.12 6.08 1.41 1.46 1.12 .07 ¼ .937
Judgement 6.15 5.95 6.17 1.49 1.66 1.20 .47 ¼ .627
Benign disinhibition 2.86 2.89 2.93 .64 .62 .49 .40 ¼ .673
Toxic disinhibition 1.62 1.63 1.56 .89 .86 .78 F (.26) ¼ .769

Note: Since our data did not meet the homogeneity of variances assumption (except for helping behavior in scenario B and toxic disinhibition), we adopted Welch test
and Games-Howell method to perform the analyses.

Table 4
Correlation among helping behavior, perception of seriousness, judgement, disinhibition effects and cyberbullying experience across three scenarios.
1 2 3 4 5 6 7 8 9 10 11 12 13 14

Scenario A Cyberbullying1 1
Harassment Helping behavior2 -.42** 1
Severity3 -.59** .49** 1
Judgement4 -.48** .49** .70** 1
Scenario B Cyberbullying5 .70** -.36** -.56** -.55** 1
Denigration Helping behavior6 -.36** .73** .40** .39** -.34** 1
Severity7 -.48** .46** .59** .49** -.55** .57** 1
Judgement8 -.50** .43** .62** .63** -.59** .47** .66** 1
Scenario C Cyberbullying9 .63** -.39** -.48** -.45** .72** -.37** -.57** -.51** 1
Exclusion Helping -.43** .68** .49** .44** -.47** .65** .55** .51** -.54** 1
behavior10
Severity11 -.47** .43** .58** .48** -.47** .40** .62** .52** -.60** .62** 1
Judgement12 -.44** .41** .45** .48** -.40** .37** .47** .48** -.54** .52** .65** 1
Online Benign13 -.10* .13* .01 .02 -.01 .14** .07 .02 -.11* .13* .13* .17** 1
disinhibition Toxic14 .37** -.10* -.35** -.33** .41** -.06 -.25** -.32** .23** -.21** -.23** -.18** .19** 1

4.2. The relationships between cyberbullying experience, helping 4.3. The predictive power of the assessed variables for cyberbullying
behavior, perceived seriousness, judgment and online disinhibition behavior

Table 4 reveals that both cyberbullying experience and helping Multiple regression analyses demonstrated that helping behavior,
behavior in the three scenarios were closely connected with other var­ perception of seriousness, overall judgment and two forms of online
iables. The three forms of cyberbullying experience and toxic disinhi­ disinhibitions jointly predicted students’ cyberbullying experiences
bition were significantly positively associated with each other, and across the scenarios (F ¼ 58.55–64.78, ps < .001) with an overall
cyberbullying experience was almost significantly negatively associated explanatory power ranging from 41% to 44%, and perception of seri­
with the remaining variables. Only denigration and benign disinhibition ousness and toxic disinhibition were common predictors (see Table 5).
were not significantly negatively associated with each other. In terms of the separate scenario model, helping behavior (β ¼ .17),
All helping behaviors were significantly negatively correlated with perception of seriousness (β ¼ .39) and both forms of online disinhi­
the three forms of cyberbullying experience and toxic disinhibition. bitions (βbenign ¼ .11, βtoxic ¼ .22) were good predictors of harassment.
Only helping behavior in denigration events and toxic disinhibition Perception of seriousness (β ¼ .26), overall judgment (β ¼ .33) and
were not significantly negatively associated with each other. All helping toxic disinhibition (β ¼ .24) were good predictors of denigration.
behaviors were significantly positively correlated with the remaining Helping behavior (β ¼ .23), perception of seriousness (β ¼ .31),
variables. overall judgment (β ¼ .20) and toxic disinhibition (β ¼ .09) were good
predictors of exclusion.

5
C.L. Huang et al. Computers in Human Behavior 110 (2020) 106338

Table 5
The regression analysis of research variables of three types cyberbullying experience.
Scenario A Scenario B Scenario C

DV: Bullying experience DV: Bullying experience DV: Bullying experience

B Beta t p B Beta t p B Beta t p

Helping behavior -.16 -.17 3.69 <.001 -.02 -.02 -.53 ¼ .600 -.20 -.23 4.65 <.001
Perception of seriousness -.33 -.39 7.07 <.001 -.23 -.26 4.76 <.001 -.28 -.31 5.68 <.001
Overall judgement -.05 -.05 -.91 ¼ .366 -.30 -.33 6.38 <.001 -.17 -.20 4.02 <.001
Online disinhibition
Benign disinhibition -.23 -.11 2.93 ¼ .004 -.05 -.02 -.59 ¼ .558 -.05 -.02 -.61 ¼ .540
Toxic disinhibition .30 .22 5.35 <.001 .33 .24 6.06 <.001 .12 .09 2.16 ¼ .031
2 2 2 2 2 2
R ¼ .65, R ¼ .42, △R ¼ .41, F ¼ 58.55, p < R ¼ .67, R ¼ .44, △R ¼ .44, F ¼ 64.78, p < R ¼ .66, R ¼ .44, △R ¼ .43, F ¼ 64.36, p <
.001 .001 .001

5. Discussion intentions to intervene in bullying was consistent with the past literature
(e.g., Cleemput et al., 2014). However, our finding also indicated that no
The average score of bullying in this study is low, suggesting that difference in educational stage existed in the intention to intervene in
cyberbullying was not a prevalent phenomenon among our Chinese harassment events. This finding that younger students inclined to pro­
participants. Unlike traditional bullying, individuals cannot avoid tect victims should be interpreted with caution.
cyberbullying attacks by avoiding contact with a bully online. The harm The correlation analysis showed that almost all variables were
caused by cyberbullying to victims is more serious than the harm caused closely connected with each other, indicating that exploring the rela­
by traditional bullying in terms of depth and breadth. A systematic re­ tionship between them has not only academic value but also statistical
view and meta-analysis of cyber-victimization and educational out­ value.
comes for adolescents conducted by Gardella, Fisher, and Teurbe-Tolon Finally, the regression results suggested that perception of severity
(2017) indicated that cyber-victimization is related to school attendance and toxic disinhibition were better predictors of the three types of
problems and low academic achievement. Other empirical studies cyberbullying experience, implying that the influences of these two
revealed that victims have a high degree of negative psychological traits, variables were cross-situational. Toxic disinhibition can directly posi­
including depression (Chu, Fan, Liu, & Zhou, 2018; Fahy et al., 2016) tively predict an individual’s bullying experience, which means that
and anxiety (Chu et al., 2018), and may be prone to suicidal intentions even under the real-name system, the online disinhibition effect still
or behavior (Brailovskaia, Teismann, & Margraf, 2018; Nikolaou, 2017; works. Specifically, when an individual perceives a higher degree of
Schenk & Fremouw, 2012). Therefore, despite the low average score in toxic disinhibition, he or she is more likely to engage in bullying
this study, we should not overlook cyberbullying. behavior, echoing the studies of Charaschanya and Blauw (2017) and
The finding that Chinese students have low involvement in cyber­ Udris (2014). As previous studies have pointed out, both prosocial and
bullying activities may be related to the use of the real-name system in antisocial behaviors are subject to normative expectations and con­
Chinese networks. Under this system, users must register their real straints (e.g., McKeown & Taylor, 2018). Toxic disinhibition is the
names and related information on an Internet service platform. common predictor across all scenarios, which is not surprising. In­
Although they can choose to hide their personal information and use a dividuals who are less helpful to others may be less affected by external
nickname when interacting with others, there is still a considerable norms and less inclined to help people. Therefore, when they perceive a
degree of prescriptive norms for deviant behavior. higher degree of toxic disinhibition, they are more likely to engage in
The result of gender differences in bullying experience was consis­ bullying. However, many factors affect individual behavior, and
tent with some previous studies (e.g., Barlett & Coyne, 2014; whether toxic disinhibition is a key determinant of deviation behavior
Lapidot-Lefler & Dolev-Cohen, 2015; Lee & Shin, 2017; Li et al., 2018), remains to be examined in more detail.
which indicate that boys are generally more likely than girls to bully In addition, helping behavior can negatively forecast students’ self-
others. In this study, our male participants were more likely than fe­ reported harassment and exclusion activities, and overall judgment
males to harass, denigrate and exclude others online. This finding can negatively forecast denigration and exclusion behaviors. This in­
echoed the finding regarding online disinhibition, which demonstrated dicates that positive bystander action and judgment are protective fac­
that boys perceived toxic disinhibition more strongly than girls. Under tors for specific kinds of bullying. However, why helping behavior does
the effect of toxic disinhibition, boys may more likely interact with not predict denigration behavior deserves future research.
others in an unfriendly manner online; therefore, this result was un­
surprising. Nevertheless, students from different educational stages did 6. Conclusion
not differ much in their bullying behavior, and it is recommended that
follow-up studies be conducted. This study examined the multiple relationships between individual
As expected, girls evaluated all cyberbullying events as more serious characteristics, helping behavior, perception of cyberbullying serious­
than boys, supporting previous studies (e.g., Bastiaensens et al., 2014; ness, judgment, online disinhibition and cyberbullying experience.
Huang et al., 2019). Empathy may explain this phenomenon (Huang Specifically, we found that gender differences existed in perception of
et al., 2019). Moreover, since we also found that girls were more likely seriousness and cyberbullying experience across three scenarios. There
than boys to correctly judge the first two scenarios as bullying instances, were also gender differences in levels of toxic disinhibition. Students
another possible explanation is that a better understanding of bullying from different educational stages displayed dissimilar tendencies to
may help girls realize that the behaviors described in the scenarios were protect victims. Moreover, helping behavior, perception of seriousness,
wrong, and they may have better judgment of bullying severity. If this overall judgment and two types of online disinhibition can conjointly
inference is correct, educators should consider how to enhance male predict specific cyberbullying deeds. The results offer recommendations
students’ understanding of bullying concepts. for future studies and implications for teachers to help stop
In terms of educational stage, junior high school students tended to cyberbullying.
help victims of denigration and exclusion and had strong attitudes to­
ward harassment. The finding that younger students had higher

6
C.L. Huang et al. Computers in Human Behavior 110 (2020) 106338

6.1. Recommendations for future studies exception of gender and education stage, these two factors should be
also taken into account when designing anti-cyberbullying programs.
Since relatively few studies have examined the role of online disin­
hibition, future studies can clarify whether online disinhibition also has CRediT authorship contribution statement
a moderating effect on helping behavior or cyberbullying, and can
further scrutinize which type of disinhibition is more influential. For Chiao Ling Huang: Conceptualization, Data curation, Funding
example, studies can examine whether under the interactive effect of acquisition, Formal analysis, Methodology, Validation, Software,
toxic disinhibition, people’s perception of the normative constraints of Writing - original draft, Writing - review & editing. Sining Zhang:
society and morality will change and further discourage helping be­ Conceptualization, Investigation, Data curation, Methodology, Writing -
haviors that represent expectations of the real society, thus becoming original draft. Shu Ching Yang: Methodology, Project administration,
indifferent bystanders. Resources, Supervision, Writing - review & editing.
Second, compared with other countries, less cyberbullying research
has been conducted among Chinese students. According to the 42nd Appendix
China Internet development statistics report released by China Internet
Network Information Center (CNNIC) in 2018, the Internet penetration We asked students to indicate their views toward the events that
rate in China is 57.7% as of June 2018. With the popularity of the occurred in three scenarios, including their perceptions and actions they
Internet, bullying will become increasingly serious and worthy of would take. The content of the three cyberbullying scenarios are pre­
attention. In addition, the existing research indicates that Chinese stu­ sented in Table 1 and detailed items are provided below (see item 01 to
dents may be influenced by socialism and collectivism; they value item 06). For items on the online disinhibition scale, please see item 07
interpersonal harmony and are willing to sacrifice their individual in­ to item 17.
terests for the needs of the larger group. They may perceive bullying as
collective conduct for maintaining group conformity and as such may be 01. This is a serious matter.
more likely to be involved in socially exclusive cyberbullying under peer 02. I will report this incident to a teacher.
pressure (Chan & Wong, 2015). Therefore, we look forward to follow-up 03. I will stand up for the victim.
research that examines the cyberbullying situation in Chinese society 04. I will tell protagonist that s/he should not do that.
and proposes more appropriate countermeasures to address this 05. This is cyberbullying.
problem. 06. I have always done what the protagonist did.
Third, from the perspective of data collection and analysis, immer­ 07. It is easier to connect with others through ICTs than talking in
sive technologies and cultural algorithms provide an opportunity for person.
future study. Immersive technologies (e.g., VR) can give users a highly 08. The Internet is anonymous so it is easier for me to express my true
interactive experience that can make participants feel emotionally and feelings or thoughts.
physically engaged in the cyberbullying events and further obtain a 09. It is easier to write things online that would be hard to say in real
more reliable answer. It is suggested that future studies can employ VR life because you don’t see the other’s face.
to elicit stronger perceptual and physiological reactions from partici­ 10. It is easier to communicate online because you can reply anytime
pants. In addition, cultural algorithms have adopted both microcosmic you like.
(population space) and macroscopic (belief space) views to analyze 11. I have an image of the other person in my head when I read their
collected data. Although, to our knowledge, cultural algorithms have e-mail or messages online.
usually been used in the AI field and have not been employed in the 12. I feel like a different person online.
education field. If researchers have interests in examining the process of 13. I feel that online I can communicate on the same level with others
change in individual behavior and want to take into account the influ­ who are older or have higher status.
ence of factors from different aspects, cultural algorithms are a good 14. I don’t mind writing insulting things about others online, because
approach because it incorporates several traditional methods of it’s anonymous.
modeling cultural advancement into a common conceptual framework 15. It is easy to write insulting things online because there are no
(Reynolds, 1994) and can well-reflect societal evolutionary processes. repercussions.
Some cyberbullying studies (e.g., Ochoa et al., 2011) have started to 16. There are no rules online therefore you can do whatever you
adopt this technique, and it is deserves more attention. want.
17. Writing insulting things online is not bullying.
6.2. Implications for educators to help stop cyberbullying

Educators should cultivate students’ correct Internet use and convey References
to students that the Internet is not out of touch with reality because
Al-Rahmi, W. M., Yahaya, N., Alamri, M. M., Aljarboa, N. A., Kamin, Y. B., & Moafa, F. A.
under the influence of disinhibition, people’s words and deeds online
(2019). A model of factors affecting cyber bullying behaviors among university
may be bolder. People may be more willing to express themselves or students. IEEE Access, 7, 2978–2985. https://doi.org/10.1109/
interact more intimately with others; they may also engage in behavior ACCESS.2018.2881292.
that deviates from social expectations or constraints. Even on the Barlett, C., & Coyne, S. M. (2014). A meta-analysis of sex differences in cyber-bullying
behavior: The Moderating Role of Age. Aggressive Behavior, 40, 474–488. https://doi.
Internet, people need to pay attention to their words and deeds. In org/10.1002/ab.21555.
addition, students may not know what to do when they observe a Barlinska, J., Szuster, A., & Winiewski, M. (2013). Cyberbullying among adolescent
classmate being bullied or when they experience victimization them­ bystanders: Role of the communication medium, form of violence, and empathy.
Journal of Community & Applied Social Psychology, 23, 37–51. https://doi.org/
selves. Educators should facilitate classroom discussions and activities to 10.1002/casp.2137.
teach students a variety of appropriate actions that they can take when Bastiaensens, S., Vandebosch, H., Poels, K., Van Cleemput, K., DeSmet, A., & De
witnessing or experiencing victimization. Teachers should also Bourdeaudhuij, I. (2014). Cyberbullying on social network sites. An experimental
study into bystanders’ behavioural intentions to help the victim or reinforce the
encourage students to speak up and dare students to stop bullying, bully. Computers in Human Behavior, 31, 259–271. https://doi.org/10.1016/j.
report bullying incidents or comfort victims to reduce the impact of chb.2013.10.036.
cyberbullying. In addition, according to Al-Rahmi et al.’s (2019) model, Brailovskaia, J., Teismann, T., & Margraf, J. (2018). Cyberbullying, positive mental
health and suicide ideation/behavior. Psychiatry Research, 267, 240–242.
exposure to social media and online participation exacerbates cyber­
bullying, cyber harassment and cyber tracking. Therefore, with the

7
C.L. Huang et al. Computers in Human Behavior 110 (2020) 106338

Brochado, S., Soares, S., & Fraga, S. (2017). A scoping review on studies of cyberbullying cyberbullying victimization and perpetration. Personality and Individual Differences,
prevalence among adolescents. Trauma, Violence, & Abuse, 18(5), 523–531. 133(15), 7–12.
Brody, N., & Vangelisti, A. L. (2017). Cyberbullying: Topics, strategies, and sex Li, J., Luo, C., Lin, Y., & Shadiev, R. (2018). Exploring Chinese youth’s Internet usage and
differences. Computers in Human Behavior, 75, 739–748. https://doi.org/10.1016/j. cyberbullying behaviors and their relationship. Asia-Pacific Education Researcher, 27
chb.2017.06.020. (5), 383–394.
Cao, B., & Lin, W. Y. (2015). How do victims react to cyberbullying on social networking Liu, Z. J., & Tung, Y. K. (2018). The impact of peer relationships, conformity, and
sites? The influence of previous cyberbullying victimization experiences. Computers cognitive empathy of bystander behavioral intentions for cyberbullying. Chinese
in Human Behavior, 52, 458–465. Journal of Psychology, 60(2), 101–124.
Chan, H. C., & Wong, D. S. W. (2015). Traditional school bullying and cyberbullying in McKeown, S., & Taylor, L. K. (2018). Perceived peer and school norm effects on youth
Chinese societies: Prevalence and a review of the whole-school intervention antisocial and prosocial behaviours through intergroup contact in Northern Ireland.
approach. Aggression and Violent Behavior, 23, 98–108. British Journal of Social Psychology, 57(3), p652–665.
Charaschanya, A., & Blauw, J. (2017). A study of the direct and indirect relationship Nikolaou, D. (2017). Does cyberbullying impact youth suicidal behaviors? Journal of
between online disinhibition and depression and stress being mediated by the Health Economics, 56, 30–46.
frequency of cyberbullying from victim and perpetrator perspectives. Scholar: Human Obermaier, M., Fawzi, N., & Koch, T. (2016). Bystanding or standing by? How the
Sciences, 9(2), 275–301. number of bystanders affects the intention to intervene in cyberbullying. New Media
China Internet network information center, 42nd China Internet development statistics & Society, 18(8), 1491–1507. https://doi.org/10.1177/1461444814563519.
report. Retrieved from http://www.cac.gov.cn/2018-08/20/c_1123296882.htm, Ochoa, A., Ponce, J., Jaramillo, R., Ornelas, F., Hern� andez, A., Azpeitia, D., et al. (2011).
(2018). Analysis of Cyber-bullying in a virtual social networking. In Proceedings of 2011 11th
Chu, X. W., Fan, C. Y., Liu, Q. Q., & Zhou, Z. K. (2018). Cyberbullying victimization and international conference on hybrid intelligent systems (pp. 229–234). HIS. https://doi.
symptoms of depression and anxiety among Chinese adolescents: Examining org/10.1109/HIS.2011.6122110.
hopelessness as a mediator and self-compassion as a moderator. Computers in Human Quirk, R., & Campbell, M. (2015). On standby? A comparison of online and offline
Behavior, 86, 377–386. witnesses to bullying and their bystander behaviour. Educational Psychology, 35(4),
Cleemput, K. V., Vandebosch, H., & Pabian, S. (2014). Personal characteristics and 430–448.
contextual factors that determine "helping. Joining in,“ and ”doing Nothing” When Resett, S., & Gamez-Guadix, M. (2017). Traditional bullying and cyberbullying:
Witnessing Cyberbullying. Aggressive Behavior, 40(5), 383–396. https://doi.org/ Differences in emotional problems, and personality. Are cyberbullies more
10.1002/ab.21534. Machiavellians? Journal of Adolescence, 61, 113–116. https://doi.org/10.1016/j.
DeSmet, A., Bastiaensens, S., Van Cleemput, K., Poels, K., Vandebosch, H., Cardon, G., adolescence.2017.09.013.
et al. (2016). Deciding whether to look after them, to like it, or leave it: A Reynolds, R. G. (1994). An introduction to cultural algorithms (pp. 131–139). San Diego,
multidimensional analysis of predictors of positive and negative bystander behavior CA: World Scientific Press.
in cyberbullying among adolescents. Computers in Human Behavior, 57, 398–415. Salmivalli, C., Lagerspetz, K., Bj€ orkqvist, K., Osterman,
€ K., & Kaukiainen, A. (1996).
Fahy, A. E., Stansfeld, S. A., Smuk, M., Smith, N. R., Cummins, S., & Clark, C. (2016). Bullying as a group process: Participant roles and their relations to social status
Longitudinal associations between cyberbullying involvement and adolescent mental within the group. Aggressive Behavior, 22, 1–15.
health. Journal of Adolescent Health Official Publication of the Society for Adolescent Schacter, H. L., Greenberg, S., & Juvonen, J. (2016). Who’s to blame?: The effects of
Medicine, 59(5), 502–509. https://doi.org/10.1016/j.jadohealth.2016.06.006. victim disclosure on bystander reactions to cyberbullying. Computers in Human
Ferreira, P. C., Simao, A. M. V., Ferreira, A., Souza, S., & Francisco, S. (2016). Student Behavior, 57, 115–121.
bystander behavior and cultural issues in cyberbullying: When actions speak louder Schenk, A. M., & Fremouw, W. J. (2012). Prevalence, psychological impact, and coping
than words. Computers in Human Behavior, 60, 301–311. of cyberbully victims among college students. Journal of School Violence, 11(1),
Fowler, F. J., Jr. (1988). Survey research methods (Rev. ed.). Thousand Oaks, CA: Sage. 21–37. https://doi.org/10.1080/15388220.2011.630310.
Gardella, J. H., Fisher, B. W., & Teurbe-Tolon, A. R. (2017). A systematic review and Song, J., & Oh, I. (2018). Factors influencing bystanders’ behavioral reactions in
meta-analysis of cyber-victimization and educational outcomes for adolescents. cyberbullying situations. Computers in Human Behavior, 78, 273–282. https://doi.
Review of Educational Research, 87(2), 283–308. https://doi.org/10.3102/ org/10.1016/j.chb.2017.10.008.
0034654316689136. Suler, J. (2004). The online disinhibition effect. CyberPsychology and Behavior, 7(3),
Huang, C. L., Yang, S. C., & Hsieh, L. S. (2019). The cyberbullying behavior of 321–326. https://doi.org/10.1089/1094931041291295.
adolescents in an online gaming environment. Children and Youth Services Review, Udris, R. (2014). Cyberbullying among high school students in Japan: Development and
106, 1–10. validation of the Online Disinhibition Scale. Computers in Human Behavior, 41,
Joinson, A. (1998). Causes and implications of disinhibited behavior on the internet. 253–261. https://doi.org/10.1016/j.chb.2014.09.036.
Revista Mexicana de Física, 10, 43–60. Willard, N. (2003). Off-campus, harmful online student speech. Journal of School
Koehler, C., & Weber, M. (2018). Do I really need to help?!” Perceived severity of Violence, 2(1), 65–93. https://doi.org/10.1300/J202v02n01_04.
cyberbullying, victim blaming, and bystanders’ willingness to help the victim. Willard, N. (2007). Cyberbullying and cyberthreats: Responding to the challenge of online
Cyberpsychology: Journal of Psychosocial Research on Cyberspace, 12(4). https://doi. social cruelty, threats, and distress. Champaign, IL: Research Press.
org/10.5817/CP2018-4-4. article 4. Wong-Lo, M., & Bullock, L. M. (2014). Digital metamorphosis: Examination of the
Kowalski, R. M., & Limber, S. P. (2007). Electronic bullying among middle school bystander culture in cyberbullying. Aggression and Violent Behavior, 19, 418–422.
students. Journal of Adolescent Health, 41, 22–30. https://doi.org/10.1016/j. https://doi.org/10.1016/j.avb.2014.06.007.
jadohealth.2007.08.017. Wright, M. F., Harper, B. D., & Wachs, S. (2019). The associations between cyberbullying
Lapidot-Lefler, N., & Dolev-Cohen, M. (2015). Comparing cyberbullying and school and callous-unemotional traits among adolescents: The moderating effect of online
bullying among school students: Prevalence, gender, and grade level differences. disinhibition. Personality and Individual Differences, 140(1), 41–45.
Social Psychology of Education, 18(1), 1–16. https://doi.org/10.1007/s11218-014- Yang, X., Wang, Z., Chen, H., & Liu, D. (2018). Cyberbullying perpetration among
9280-8. Chinese adolescents: The role of interparental conflict, moral disengagement, and
Lee, C., & Shin, N. (2017). Prevalence of cyberbullying and predictors of cyberbullying moral identity. Children and Youth Services Review, 86, 256–263.
perpetration among Korean adolescents. Computers in Human Behavior, 68, 352–358. Zhou, Z., Tang, H., Tian, Y., Wei, H., Zhang, F., & Morrison, C. M. (2013). Cyberbullying
Leung, A. N. M., Wong, N., & Farver, J. M. (2018). Cyberbullying in Hong Kong Chinese and its risk factors among Chinese high school students. School Psychology
students: Life satisfaction, and the moderating role of friendship qualities on International, 34(6), 630–647. https://doi.org/10.1177/0143034313479692.