Abstract
Cyberbullying has surfaced as a worldwide concern in recent decades. As the largest online community globally, China has experienced a spike in cyberbullying across social groups and functions. Using survey data collected from roughly 1,600 students from two universities, this study assesses students’ intervention behaviours toward cyberbullying. We assess the connections between students’ social bonds and the perceived severity of cyberbullying and their intention to stop, do nothing, or even support when witnessing cyberbullying. We found that individuals with stronger social bonds are more willing to ask perpetrators to stop bullying. The positive link between social control and intervention is stronger among those who reported a higher sense of powerlessness. In addition, respondents who have stronger social bonds and who consider cyberbullying as more serious have lower tendencies to keep neutral. The negative connection between social bonds and doing nothing is stronger among students with higher senses of powerlessness. Implications for policy and future research are discussed.
Introduction
Six months after sharing a photograph on Chinese social media celebrating her graduate school admission with her bedridden grandfather, Zheng Linghua died. The 23-year-old had shared the image on the platform Xiaohongshu – sporting pink hair and visibly excited, she announced to the world that she had earned a spot to study music at East China Normal University… But her joy was short-lived. Within days, she had become a target of bullying online. Her photograph was shared with false, and often insulting, captions. She then became the target of relentless taunting, with some calling her a ‘nightclub girl’ and an ‘evil spirit’. (Ng, 2023, March 26)
Cyberbullying has surfaced as a worldwide concern over the past two decades. As the largest online community globally, with over one billion netizens in 2023 (Xinhua News, 2024), China has experienced a spike in cyberbullying across social groups and functions (Han et al., 2021; Li et al., 2023; Xing et al., 2023). The vicious trolling leading to Ms. Zheng's tragedy, unfortunately, does not come as a surprise, as cyber victimisation has been linked to suicidality in China (Meng et al., 2023) and elsewhere (Hinduja & Patchin, 2019; Strohacker et al., 2021). A recent national survey of more than 2,000 young (aged 18–35) Chinese social media users found that nearly 40% have experienced some form of online abuse, with 16% of the victims reporting suicidal thoughts (WHYNOT, 2022). The rise of cyberbullying worldwide has stirred a large body of research on various correlates of cyberbullying penetration and victimisation in China (e.g., Ding et al., 2024; Su et al., 2022; Wang et al., 2021; Yang et al., 2022). Nonetheless, several under-researched areas remain to be filled.
This study assesses factors related to young people's willingness to stop, ignore or reinforce cyberbullying. We examined whether social bonds and a sense of powerlessness are connected to students’ intention to intervene, stay neutral and support such interpersonal violence. This study expands the current literature on cyberbullying on several fronts. First, although the past decade has witnessed an explosion of research on many aspects of cyberbullying in China, very little scholarly work has investigated people's intention to intervene in cyberbullying incidents (see Dang & Liu, 2022 for a notable exception). Our study conceptually classifies college students’ reactions to cyberbullying into three forms, including stopping cyberbullying (e.g., persuading the perpetrators to stop their action), doing nothing (e.g., remaining neutral and not taking any action) and supporting cyberbullying (e.g., encouraging the perpetrators through likes and supportive comments). Our findings would shed light on the different and similar predictors of these various forms of responses.
Second, despite being one of the most tested criminological theories, the social bonds theory (Hirschi, 1969) has not been utilised much in the cyberbullying literature (see Costello et al., 2022; Guo, 2022). To address the lack of diversity in theoretical and empirical applications, our study represents a first attempt to assess whether young people's social bonds may be linked to their willingness to intervene in cyberbullying. It extends the usage of the social bonds theory to explaining not only criminal behaviours but also bystanders’ behaviours. Finally, we also included the pervasive sense of powerlessness among young Chinese people (Jin et al., 2019; UCLA Center for Chinese Studies, 2023) as a key correlate of responses to cyberbullying situation. When young people feel that they are unable to stop cyberbullying and/or are afraid of being a victim of cyberviolence themselves, they are less likely to intervene actively. Indeed, the rational choice theory would argue that when the perceived benefits (e.g., the likelihood of successfully stopping a cyberbullying incident) cannot outweigh perceived costs (e.g., the fear of being victimised by cyberbullying), people will not step up to stop cyberbullying. In addition, perceived powerlessness may further moderate the association between social bonds and intervention, perhaps with the positive influence of social bonds on intervention getting weaker when people feel more powerless. This study allows us to explore the nuanced linkages between important social phenomena of social bonds, perception of powerlessness and willingness to intervene in cyberbullying through the recounts of young Chinese.
Drawing upon survey data from over 1,600 students in two Chinese universities, this study is designed to address two research questions: (1) Are Chinese college students’ intervention behaviours connected to their social bonds and a sense of powerlessness, after controlling for a wide range of variables including perceived severity of cyberbullying, and (2) Are the relationships between social bonds and intervention moderated by the sense of powerlessness? We expect our findings to bear important implications for policy and future research on cyberbullying.
Literature review
Bystander responses to cyberbullying
A line of inquiry has analysed bystanders’ intervention behaviours when facing cyberbullying. Pöyhönen et al. (2012) classified the bystanders’ roles in cyberbullying into three common types: defenders, reinforcers and passive responders, with the first group defending the victims, the second group reinforcing the bullying, and the last group staying neutral or doing nothing. Based on adolescent data from South Korea, a recent study expanded the bystander typology into five groups: limited bystanders, pro-bullies, outsiders, defenders and inconsistent bystanders (Hong & Lee, 2022), showing the complexity and diversity of the bystanders’ roles in cyberbullying. Following Bastiaensens and co-researchers’ work (2014), most recent studies focused on bystanders’ behaviours of supporting the victim and reinforcing the cyberbully (e.g., Hayashi & Tahmasbi, 2022; Herry & Mulvey, 2023; Olenik-Shemesh et al., 2017; Rebollo-Catalan & Mayor-Buzon, 2020). A few studies touched on the passive behaviour of doing nothing (Machackova et al., 2018) and victim blaming (Bedrosova et al., 2023).
Studies have also reported the prevalence of bystanders’ intervention behaviours. Henson et al. (2020) found that the most common form of intervention involved discussing the possible dangers of risky online behaviours with others (53%–58% of the respondents), followed by attempting to help or get help for victims (34%–55%) and having confrontation (i.e., speaking out) with the perpetrator (28%–41%). Rebollo-Catalan and Mayor-Buzon's study (2020) showed that the most common bystander intervention was helping the victim (64.5%), followed by doing nothing (31.5%), helping the aggressor (2.1%) and encouraging the aggressor (1.5%). Although past studies have revealed a higher prevalence rate for favourable (e.g., supporting the victim) than unfavourable actions (e.g., reinforcing the bullying), the likelihood of bystanders engaging in various responses to cyberbullying could vary by demographical and situational characteristics. For instance, gender differences were detected in intervention behaviours, with women more likely to support the victim and less likely to support cyberbullying than men (Herry & Mulvey, 2023; Olenik-Shemesh et al., 2017; Rebollo-Catalan & Mayor-Buzon, 2020). Also, past studies linked the seriousness of cyberbullying and the number of bystanders to bystanders’ intervention behaviours, with the levels of seriousness and number of bystanders positively relating to positive interventions (Bastiaensens et al., 2014; Patterson et al., 2017).
In China, a few studies have assessed bystanders’ behaviours toward cyberbullying. Dang and Liu (2022) investigated the relationships between high school students’ pro-cyberbullying personal and social norms and two behavioural responses, helping the victim and reinforcing the cyberbullying. They found that pro-cyberbullying personal and social norms were negatively related to the intentions to help the victim. Moreover, higher pro-cyberbullying personal norms were correlated with higher intentions to reinforce the bully when pro-cyberbullying social norms were high yet had no such effect on reinforcing when pro-cyberbullying social norms were low. Diverging from Dang and Liu (2022), we focused on social bonds instead of social learning (e.g., social norms on cyberbullying), examining the independent as well as the interactional effects of social bonds and senses of powerlessness on three responses to cyberbullying incidents. In the next section, we elaborate on these concepts and their associations with cyberbullying intervention actions.
Social bonds and sense of powerlessness
Hirschi's (1969) seminal work posited that delinquency and crime would likely occur when individuals’ bonds to conventional society were weakened. Four key concepts reflect Hirschi's social bonds: attachment, commitment, involvement and belief. Attachment focuses on affection toward significant people in life, especially parents for children and juveniles, with higher attachment curbing delinquency. Commitment signals people's devotion to conventional forms of behaviour such as school and work, with greater commitment inhibiting delinquent acts. Involvement refers to participating in conventional activities, allowing people less time to engage in crime and delinquency. Beliefs reflect people's agreement with commonly shared conventional values and norms, with stronger belief reducing delinquent and criminal behaviour.
Although many criminological studies have tested the social bonds-delinquency nexus, less than a handful of studies have drawn upon the social bonds theory to investigate correlates between cyberbullying offending and victimisation. Using U.S. adolescents between the ages of 15 and 18 as the study sample, Costello and colleagues (2022) found that closeness to family reduced the likelihood of joining an ongoing cyberattack, whereas closeness to online communities increased such probability. Guo's (2022) study showed that parental and school attachment has a protective effect against delinquency among a sample of U.S. adolescents. Specifically, perceived bonds from parents, friends and teachers reduced the risk of physical fighting among cyberbullying victims, and perceived bonds from friends and teachers reduced the risk of physical fighting among cyber bullies.
Some studies incorporated social bond elements, particularly attachment to parents and schools, into the research on cyberbullying without necessarily mentioning the theory. For instance, a review of 25 articles on cyberbullying revealed that perceived supportive relationships with parents, teachers and schools increased the likelihood of defending the victim (Lambe et al., 2019). Acosta and collaborators’ (2019) study showed that school connectedness mediated the relationships between school climate (e.g., teacher support, peer interactions and student input) and cyberbullying among U.S. middle school students. A review of Korean cyberbullying literature revealed that a lack of parental attachment predicted cyberbullying perpetration (Lee et al., 2018). A few studies conducted in Chinese societies also demonstrated that parental attachment mediated the connections between cyberbullying victimisation and youth health problems (Zhu et al., 2022) and that negative school experiences and school involvement were associated with adopting an avoiding style to cope with cyber victimisation among adolescents (Chan & Wong, 2017).
In addition to social bonds’ influence on perpetrators and victims, this study aims to explore the relationship between social bonds and different responses to cyberbullying. The linkage between social bonds and bystanders’ attitudes and behaviours toward cyberbullying may be accounted for by at least three mechanisms. First, social bonds, especially parental bonds and care, can affect individuals’ character strengths in areas that relate directly to their willingness to intervene in cyberbullying and stop wrongdoing. These strengths include bravery, empathy, kindness and a sense of fairness (Nadler, 2018; Ngai, 2015). Second, acts of bravery when facing illegal or immoral acts are often outward manifestations of a person's internal sense of morality (Covington, 2020). Intervening in cyberbullying reflects an endeavour to uphold one's internal moral order, which reflects general conventional social norms. As such, individuals who hold stronger conventional beliefs are expected to be more likely to actively intervene in cyberbullying incidents. Lastly, social bonds, which facilitate social integration and accumulate social capital, may also promote civil courage of intervention in wrongdoing by alleviating any concerns about potential risks and consequences of the intervention (Kühnlenz & Kühnlenz, 2020).
Although the power imbalance between perpetrators and victims has been considered an important correlate of cyberbullying and cyber victimisation (Barlett et al., 2022; Wang & Ngai, 2022), little is known about how people's sense of powerlessness is related to their choices of cyberbullying intervention. Young Chinese people have been increasingly feeling powerless in recent years, resembling their counterparts before the 1980s when the country started its economic reforms (UCLA Center for Chinese Studies, 2023). A qualitative study identified three dispositions of online Chinese participants in social and political events: angry youth, powerless cynics and realistic idealists (Fu, 2021). The powerless cynics expressed their pessimistic views about making any desired changes in their social and political lives and showed disillusionment with the power of social media (Fu, 2021). The recent movements of “lying flat” (tang ping) and “tanking” (bai lan) have crystalised young Chinese’ feelings of powerlessness and discontent with social and political environments (Yeh, 2022). Researchers also found that people's sense of powerlessness may support the existing political and social structures and institutions that disadvantage them (Van der Toorn et al., 2015). Extending these arguments to online offending, one may speculate that individuals’ sense of powerfulness when it comes to cyberbullying makes them approach and respond to cyberbullying by keeping their hands-off or even reinforcing or supporting such situations.
Besides testing the connections between the sense of powerlessness and reactions to cyberbullying, this study also proposes that the sense of powerlessness can moderate the relationships between social bonds and responses to cyberbullying. In other words, people's degree of feeling powerless when it comes to cyberbullying may shape the nature of the relationship between their social bonds and reactions to cyberbullying. For instance, we would expect a positive linkage between social bonds and willingness to intervene in cyberbullying, but such an association could be stronger among young people with a lower sense of powerlessness. In other words, the effects of social bonds and feelings of adequacy may be reinforcing each other in shaping intervention intentions. Similarly, one would suspect a negative link between social bonds and no intervention or unfavourable intention (e.g., supporting cyberbullying); such a connection may be stronger among young people with a greater sense of powerlessness.
The current study
This study assessed important correlates of different modes of response to cyberbullying situations. Although cyberbullying has drawn much research attention in China and worldwide, a few areas remain under-researched. We anchored our study on reaction to cyberbullying through the lens of the criminological theory of social bonds and the recent trend of a growing sense of powerlessness in social and economic lives including when witnessing cyberbullying among young people in China, while taking other more traditional factors such as the perception of the seriousness of cyberbullying into account.
Based on the literature review and our research questions, the following hypotheses were formulated and tested in this study. H1a: College students with stronger social bonds are more likely to stop cyberbullying. H1b: College students with stronger senses of powerlessness are less likely to stop cyberbullying. H1c: The positive connection between social bonds and stopping intervention is stronger among college students with lower senses of powerlessness. H2a: College students with stronger social bonds are less likely to do nothing when facing cyberbullying. H2b: College students with stronger senses of powerlessness are more likely to do nothing when witnessing cyberbullying. H2c: The negative connection between social bonds and doing nothing is stronger among college students with lower senses of powerlessness. H3a: College students with stronger social bonds are less likely to support cyberbullying. H3b: College students with stronger senses of powerlessness are less likely to support cyberbullying. H3c: The negative connection between social bonds and support for cyberbullying is stronger among college students with lower senses of powerlessness.
Method
Data collection and sample
This study analysed survey data collected from Chinese college students by a research project initiated by a group of China and United States-based scholars in the fall of 2023. The project investigates factors related to Chinese college students’ online behaviours and attitudes toward online violence. Two Chinese universities in two provinces were selected as study sites due mainly to the researchers’ connections to faculty in both institutions, who were willing to assist with data collection. With over 30,000 students, the first university (University A) is in a coastal province known for its rich political and cultural traditions and advanced economic developments. Belonging to the list of China's “Double First-Class” universities, University A is a comprehensive higher education institution focusing on engineering and science disciplines. The second university (University B) has approximately 17,000 students and is situated in a non-coastal, eastern Chinese province with moderate economic development. It is a general undergraduate university with an emphasis on engineering majors. The dual-site design allowed the researchers to analyse whether varying school types, sizes and locations may be connected to different online attitudes and behaviours.
The research team developed a survey questionnaire in 2023. In early October 2023, a small group of Chinese students from two universities participated in the survey's pre-test. Based on the feedback from the pre-test, several minor modifications were made to the survey instrument to improve the understandability of all items for college students. Once finalised, the instrument was built on a popular online survey tool.
Data collection started at the two universities in late October 2023 and lasted for about one month. Purposive and convenience sampling methods were employed, targeting students enrolled in required courses. Project researchers identified required courses across different college years and majors and contacted course instructors for their permission to conduct surveys with their students. Once permission was granted, researchers visited the courses at scheduled times to inform all students who were present in the classrooms about the study's purpose and voluntary and anonymous nature. Students were also assured that non-participation would not adversely affect their performance in these classes. With the help of course instructors, the survey link was sent to class WeChat groups, and students completed the survey in the classroom using their cell phones. WeChat is the largest social media app in China widely used by government employees and the public. The average time for students to complete the survey was 15‒20 min.
During data collection, the survey was distributed to roughly 1,200 and 1,500 students in universities A and B, respectively. Among them, 832 students in University A and 1,098 students in University B completed the survey, generating an initial sample of 1,930 students. We dropped 63 cases from University A and 240 cases from University B due to unreasonably short response time (i.e., less than three minutes), straight-lining (i.e., giving identical answers to items) and missing values on variables, resulting in a final study sample of 1,627 students, including 769 students from University A and 858 from University B. As shown in Table 1, the gender and major distributions of the study sample elucidate both universities’ emphasis on engineering and sciences. More than 70% of the sample students are males, and 67% have engineering majors. The average age of the sample students is 19, and about two-thirds have a rural/agricultural hukou (household registration status) and affiliation to the communist party.
Descriptive statistics of study variables (N = 1,618).
The un-centralised mean for social bonds and powerlessness is 36.65 and 12.59, respectively.
Variables
This study used three dependent variables, two independent variables, one interaction term and eight control variables. Three dependent variables were constructed based on a series of items asking the respondents’ willingness to stop, keep neutral or not intervene, and support when witnessing nine different forms of cyberviolence, including, for example, cyber harassment and bullying and researching and publicly sharing personal identifiable data (Brody, 2013). The measurement of cyberbullying types was informed by the categorisations of online aggression as delineated by Shapka et al., encompassing forms such as Direct Cyberbullying, Indirect Cyberbullying, Social Exclusion, and Cyber Invasion of Privacy. Our assessment instrument comprised nine questions that served as metrics for these types. Appendix 1 displays a complete list of these nine forms of cyberbullying. Response categories ranged from “definitely not” (=1) to “definitely will” (=6). We labelled the dependent variables as “stop cyberbullying”, “doing nothing”, and “support cyberbullying”. All three variables were measured as additive scales with high internal consistency (see Table 1). Before constructing the three outcome variables, Exploratory Factor Analysis (EFA) that specifies one-factor, two-factor and three-factor solutions among all indicators of the three outcomes were performed. The EFA results, as displayed in Appendix 2, indicated that a three-factor solution fits the data the best. The one-factor and two-factor solutions had poor model fit.
Three independent variables include social bonds, a sense of powerlessness, and their interaction term. The measurement of social bonds in our study was anchored in Hirschi's (1969) conceptualisation of four fundamental elements of social bonding: attachment, commitment, involvement, and belief. Drawing on measurement indices presented by Hirschi (2004) and Chapple et al. (2005), our scale included items pertaining to parental attachment, school commitment, and additionally incorporated indicators of engagement in volunteer activities and adherence to conventional moral beliefs. This construct was operationalised through nine questions that served as metrics for these dimensions. Sense of powerlessness was constructed based on three items, signalling the respondents’ feelings of being unable to stop cyberviolence and afraid of being a victim of cyberviolence. Items used for both scales have response categories varying from strongly disagree (=1) to strongly agree (=6). An interaction variable between social bonds and a sense of powerlessness was constructed by multiplying the two scales (after centering). All survey items employed to construct the three independent variables are listed in Appendix 1. The Cronbach alpha associated with these scales suggests high internal reliability.
Eight variables were controlled in data analysis. The first control variable, perceived seriousness of cyberbullying, is a nine-item scale, asking respondents to rate the seriousness (not at all serious = 1; very serious = 6) of the same nine forms of violence used to construct the three dependent variables. Gender, hukou (i.e., household registration), college major, political party affiliation, and school are dummy variables with one (1) representing female, agricultural hukou, engineering major, communist party members and University A, respectively. Age was measured in years. Perceived social class is five-category variable, ranging from lower class (=1) to top class (=5). Table 1 demonstrates the descriptive statistics for variables used in the study.
Analytic plan
Data analysis was conducted using Mplus 8.3. Ordinary least-squared regression was used to investigate the effects of all independent variables on the continuous outcome variables. The Johnson-Neyman (JN) plot is utilised to illustrate the moderating effect of a sense of powerlessness on the relationship between social bonds and responses to cyberbullying. The JN technique is generally considered a more statistically rigorous and more informative alternative to simple slope analysis for exploring interactions of two continuous variables (Preacher et al., 2007).
Results
Table 2 displays the regression results from three models. For the stop cyberbullying model, individuals with higher levels of social bonds are more willing to intervene, which is consistent with our hypothesis (H1a). It means that for each unit increase in social bonds, there is, on average, a 0.37-unit increase in the tendency to intervene, considering all levels of the powerlessness variable. Disconfirming our hypothesis (H1b), the level of powerlessness does not have a significant connection to students’ willingness to stop cyberbullying. Noticeably, albeit contradictory to our hypothesis (H1c), a significant interactional relationship between social bonds and powerlessness on intervention is observed, with the positive link between social bonds and stop cyberbullying is stronger among those who reported higher levels of powerlessness (see Figure 1 for illustration). Among the control variables, college students’ views of cyberbullying seriousness are not predictive of their willingness to intervene. Respondents who had an agricultural hukou and who reported a higher family socioeconomic status were more willing to stop cyberbullying, whereas students from University A were less willing to intervene.
The positive effect of social bonds on stop cyberbullying increases by the level of sense of powerlessness.
Ordinary least squares regression results (N = 1,618).
*p < .05; **p < .01; ***p < .001.
Looking at the second model of doing nothing, respondents with stronger social bonds have lower tendencies to keep neutral or do nothing, confirming our hypotheses (H2a). It means that for each unit increase in social bonds, there is, on average, a 0.21-unit decrease in the tendency to do nothing, considering all levels of the powerlessness variable. Also supporting our hypothesis (H2b), those who think they are more powerless are more inclined not to intervene. Inconsistent with our hypothesis (H2c), the interactional relationship between social bonds and powerlessness on doing nothing is significant and negative, suggesting that the negative connection between social bonds and being neutral is stronger among those with higher senses of powerlessness (see Figure 2). This interaction effect echoes the above result of interactional relationships in the stop cyberbullying model. For the control variables, students who consider cyberbullying as more serious are less likely to do nothing, while students from the University A are more likely to stay neutral and not intervene.
The negative effect of social bonds on doing nothing amplifies by the level of sense of powerlessness.
Concerning support cyberbullying, neither social bonds nor powerlessness has any independent or interactional connection to college students’ support for cyberbullying, which is inconsistent with our hypotheses (H3a, H3b and H3c). Perceived seriousness, as expected, has a negative association with support cyberbullying, with individuals who consider cyberbullying less serious indicating more supportive attitudes toward cyberbullying. Additionally, students who are older, with an agricultural hukou, perceiving a higher family social class, and from the regional university expressed greater support for cyberbullying compared to their counterparts.
Discussion
Although the correlates of cyberbullying have been widely investigated in China, little research has focused on bystanders’ roles or intervention behaviours toward online bullying. Drawing upon survey responses from Chinese college students, our study confirmed that students’ social bonds and feelings of powerlessness about cyberviolence are connected to their self-reported reactions to online bullying. We also found interesting interactions between social bonds and the sense of powerlessness in shaping college students’ intervention behaviours. We elaborate on our main findings below.
First, social bonds matter in predicting intervention behaviours when witnessing cyberviolence. Respondents with stronger social bonds are more likely to stop cyberbullying and less likely to do nothing as bystanders. Beyond the many studies showing that social bonds worked to prevent criminal offending during emerging adulthood (Salvatore & Taniguchi, 2012), our findings indicate that such bonds could also be instrumental in fostering favourable behavioural tendencies toward bullying among college students. However, contrary to previous findings (see Costello et al., 2022), our analysis failed to establish a significant negative link between social bonds and supporting cyberbullying, suggesting that social bonds do not necessarily inhibit unfavourable bystander responses to cyberbullying. Given limited attention to social bonds in the cyberbullying literature, future research should continue utilising social bonds to investigate the correlates of bystanders’ intervention behaviours.
Second, college students’ greater sense of powerlessness is linked to a higher tendency not to do anything as witnesses of cyberbullying. Although this finding confirms our hypothesis, the absence of relevant evidence in the existing literature reminds us of the need to further explore why the sense of powerlessness is only connected to passive reactions (e.g., doing nothing) but not to active interventions (e.g., either stopping or supporting cyberbullying). Moreover, we also found that there is an interactional effect between social bonds and sense of powerlessness about cyber bullying on modes of reaction to cyberbullying. Specifically, stronger social bonds are associated with a greater likelihood of stopping bullying actions for people with higher senses of powerlessness. Similarly, the negative influence of social bonds on the tendency to do nothing when witnessing cyberviolence is stronger among people with greater senses of powerlessness. That is, the beneficial effect of social bonds on promoting positive reactions to cyberviolence incidents are especially salient among students who feel less adequate about curbing cyberbullying. It seems that for students who feel unable to stop cyberviolence and afraid of being a victim of cyberviolence, social bonds are especially important for them to be willing to step up to address such violence. Perhaps the social support and capital resulting from social bonds can provide them with much-needed strength and courage to intervene actively. These findings highlight the intertwined nature of these concepts and call for more studies using different designs, such as in-depth interviews or small focus groups, to further understand their connections.
Third, consistent with previous findings (Bastiaensens et al., 2014; DeSmet et al., 2014; Hayes, 2019), we found that the perceived seriousness of online bullying reduced the likelihood of supporting cyberbullying and doing nothing, but seriousness did not necessarily lead to greater intentions to actively intervene to stop cyberbullying among young Chinese. These results suggest that perceived seriousness and social bonds play a different role in influencing college students’ reactions to cyberbullying. Perceived seriousness curtailed support for cyberviolence, whereas social bonds stirred stopping-bullying reactions, with both lowering the possibility of doing nothing. These findings implied college students’ different considerations in deciding their courses of action toward cyberbullying, illustrating the need for future studies to include multiple types of intervention in analysing bystanders’ responses to cyberbullying.
Finally, three additional control variables, hukou (household registration), social class, and university, significantly correlate with at least two of the three response behaviours. Students with a rural/agricultural background and higher social class were more active in both stopping and supporting cyberbullying than students with an urban background and lower social class. Students at the large university were also less likely to stop and support cyberbullying and more likely to do nothing. In addition, older students were more supportive of cyberbullying than their younger counterparts. These findings demonstrate the need to control for respondents’ background characteristics when examining reactions to cyberbullying and the need to collect qualitative data to understand the reasons behind these differences across demographic and student groups.
Before discussing policy implications, a few limitations should be acknowledged. First, we utilised a non-probability-based sample of students from two universities that contain students overrepresenting engineering and sciences. We cannot ensure the study sample's representativeness and our findings’ generalisability to the student populations in these universities, not to mention other universities nationwide. Future research should increase the sample diversity by gathering information from students at more universities in different regions to further test their attitudes and behaviours pertaining to cyberviolence. Second, although our independent variables were found connected to different forms of intervention, the relatively low explanatory power associated with our regression models indicates that some theoretically relevant variables need to be included in our analytical models. For example, pro-cyberbullying or general personal and social norms predicted intervention behaviours (Dang & Liu, 2022; Palomera et al., 2021) should be considered in future research when measures are available. Finally, although we construed three different forms of bystanders’ responses to cyberbullying, other verbal or behavioural intervention types should be considered (see Hong & Lee, 2022) to expand the full range of bystander responses.
Our study findings shed light on some directions for policymakers and school administrators. Our findings indicate that college students’ social bonds, reflected in forms such as commitment to academic excellence and participation in extracurriculum activities, can foster positive interventions to stop cyberbullying and reduce staying neutral when witnessing bullying as bystanders. Effective strategies and programs aimed at helping college students achieve academic excellence should be implemented, such as strengthening mentoring and counseling systems, promoting clear and achievable academic and career goals, cultivating peer feedback and support, and building effective time management skills. Encouraging volunteer work among college students can be beneficial in curbing cyberbullying. University officials hold a key role in cultivating volunteering among students. By equipping students with the necessary information and skills needed for volunteering, illustrating the profound impact that they can have on their community and their professional development, allowing them to integrate their volunteer work with their degree requirements, and recognising and rewarding them with exemplary community service work, school officials can empower students to actively combat cyberbullying. Moreover, targeted cyberbullying prevention and reduction programs can help deter older, rural, and better-off college students from supporting cyberbullying. Most importantly, the government and educators should improve the public's awareness of the seriousness of cyberviolence, effectively control cyberviolence behaviours, provide victims with adequate support, and hold perpetrators accountable. Enhancing public understanding of nature, extent, and consequences of cyberviolence as a social problem and providing them with needed tools and resources, be it social or legal support, to fight against cyberbullying is the key to encourage positive responses from the public, including students, to stand up and fight cyberbullying.
Footnotes
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Appendix 1. Survey items used to construct perceptual scales
Stop Cyberbullying (definitely not = 1, definitely will = 6)
I will persuade the perpetrators to stop their action, when I see them engage in the following online behaviour:
Liking offensive comments about others Sharing or reposting offensive comments about others Spreading fabricated rumours Sharing or reposting unverified news reports Publicly disclosing others’ private information, such as name, identity, address, private life or whereabouts Creating web pages, videos, short essays, etc., to humiliate or defame others Initiating doxing (human flesh search) Commenting on others using insulting, cursing, sarcasming, and threatening language online Mass sending or organising netizens to collectively insult others or boycott something
Doing Nothing (definitely not = 1, definitely will = 6)
I will remain neutral and not take any action, when I see people engage in the following online behaviour:
Liking offensive comments about others Sharing or reposting offensive comments about others Spreading fabricated rumours Sharing or reposting unverified news reports Publicly disclosing others’ private information, such as name, identity, address, private life or whereabouts Creating web pages, videos, short essays, etc., to humiliate or defame others Initiating doxing (human flesh search) Commenting on others using insulting, cursing, sarcasming, and threatening language online Mass sending or organising netizens to collectively insult others or boycott something
Support cyberbullying (definitely not = 1, definitely will = 6)
I will encourage the perpetrator(s) through likes and supportive comments when I see them engage in the following online behaviour:
Liking offensive comments about others Sharing or reposting offensive comments about others Spreading fabricated rumours Sharing or reposting unverified news reports Publicly disclosing others’ private information, such as name, identity, address, private life or whereabouts Creating web pages, videos, short essays, etc., to humiliate or defame others Initiating doxing (human flesh search) Commenting on others using insulting, cursing, sarcasming and threatening language Mass sending or organising netizens to collectively insult others or boycott something I share my thoughts and feelings with my parents. When I face difficulties, I can receive emotional and financial support in real life. Getting good grades is important to me. I have clear life goals. I actively participate in volunteer activities. I have participated in volunteer work, environmental protection activities, blood donation, disaster relief and charitable donations, and other social welfare activities. I am willing to help those in need. I avoid deceitful or dishonest behaviour. The online space is very chaotic, and no organisation or individual can truly stop online violence. With limited abilities, I often feel powerless when facing with incidents of online violence. I fear being targeted on the Internet. Liking offensive comments about others Sharing or reposting offensive comments about others Spreading fabricated rumours Sharing or reposting unverified news reports Publicly disclosing others’ private information, such as name, identity, address, private life or whereabouts Creating web pages, videos, short essays, etc., to humiliate or defame others Initiating doxing (human flesh search) Commenting on others using insulting, cursing, sarcasming and threatening language online Mass sending or organising netizens to collectively insult others or boycott something
Social Bonds (strongly disagree = 1, strongly agree = 6)
Sense of Powerlessness (strongly disagree = 1, strongly agree = 6)
Seriousness of cyberbullying (not serious at all = 1, very serious = 6)
Appendix 2. Exploratory Factor Analysis (EFA) of dependent variables
| Fit statistics | Chi-Square | Df | p | CFI | TLI | RMSEA | SRMR |
|---|---|---|---|---|---|---|---|
| 1-factor | 58,277 | 324 | .000 | .325 | .268 | .332 | .368 |
| 2-factor | 29,297 | 298 | .000 | .662 | .602 | .245 | .219 |
| 3-factor | 5245 | 273 | .000 | .942 | .925 | .016 | .064 |
| Models compared | |||||||
| 1-factor vs. 2-factor | 28,979 | 26 | .000 | ||||
| 2-factor vs. 3-factor | 24,052 | 25 | .000 |