Introduction

Being online on social media has become an inseparable part of adolescents’ daily lives in most parts of the world (Boer et al., 2020). Although social media provides limitless opportunities for the adolescents, such as getting in touch with friends, expanding personal circles, and learning new skills, it has a downside when it comes to cyberbullying (Feijóo et al., 2021). Cyberbullying has been defined as repeatedly and willfully making the other person feel ashamed, hurt, and excluded through electronic use (Patchin & Hinduja, 2006), including behaviors such as sharing one’s embarrassing photos/videos beyond the original intended audience, harassing through direct messages/comments, spreading rumors, and intentionally excluding individuals from group chats. These experiences have been linked to adverse outcomes in the victims, such as depression (Gámez-Guadix et al., 2013), anxiety (Pabian et al., 2016), post-traumatic stress symptoms (Mateu et al., 2020), lowered self-esteem (Patchin & Hinduja, 2010), and lowered academic performance (Wright, 2015). However, the majority of these symptoms are present in the perpetrators of cyberbullying as well (Cañas et al., 2020; Patchin & Hinduja, 2010; Wright, 2015), showing that involvement in cyberbullying is hazardous for both the bullies’ and victims’ well-being.

Adolescents’ involvement in cyberbullying is complex. A recent meta-analysis showed that around 20 to 80% of cyberbullies were cyberbullying victims themselves (Lozano-Blasco et al., 2020). Moreover, the reported prevalence of cyberbullying varies hugely between and within countries, possibly due to the cultural influences and methods employed in the investigation. For example, 1.4 to 6.2% were cyberbullied in Korea (Lee et al., 2018), while 20.4% were cyberbullied in England (Fahy et al., 2016). Despite the differences in prevalence, the results of those studies highlight that cyberbullying has adverse effects on both perpetrators’ and victims’ well-being.

Relatively little is known about Mongolian adolescents’ experiences of cyberbullying. Mongolia is a developing country, with around 3.4 million population (Kemp, 2022). Given that the population is relatively young with the median age being 28.5 years, around 15% are adolescents aged 14 to 18 years, and 96% of those adolescents aged 13–18 years use social media (Communications and Regulatory Commissions, 2021), there is a need to investigate the cyberbullying prevalence and experiences among this population.

Adolescent social media statistics from Mongolia (Communications and Regulatory Commissions, 2021) indicate that Facebook is the most used platform (i.e., 96% of Mongolian adolescents have a Facebook account, and 71% of them use it daily), with adolescents also reporting holding accounts on YouTube (71%), Instagram (63%), and TikTok (45%). There are no gender differences in social media use, except that girls tend to use Instagram more than boys (72% vs. 54%). In terms of internet connection and data availability, as of 2022, 66% of the total population of Mongolia has a constant connection to the internet with mobile and fixed connection speeds comparable to the UK (Kemp, 2022). Therefore, for the 99% of Mongolian adolescents who report frequently connecting to the internet from home using their mobile data (Communications and Regulatory Commissions, 2021), connectivity is sufficient for regular internet usage such as social networking, streaming, and downloading. Although Mongolians interact in Mongolian, code-switching is common especially for adolescents and in the internet context, as English-taught private schools have become popular after the collapse of the Soviet Union (Marzluf, 2012).

A report from the Communications and Regulatory Commissions (2021) suggested that half of the adolescents who participated in the survey had been bullied on social media at least once. Moreover, recent baseline data on Mongolian adolescents’ emotional and behavioral problems (Optimal NMAX LLC, 2022) showed that 8.5% of the adolescents reported bullying others on the internet at some point. While there were no specific effects of age, gender differences indicated that boys tended to cyber-perpetrate two times higher than girls. This could be due to the traditional Mongolian male masculinity culture (Hurst, 2001; Rarick et al., 2014). Historically, Mongolians value masculinity and aggressive competitiveness in males and femineity and submissiveness in females, and these trends may still be present to some extent.

To develop successful anti-cyberbullying programs that are culturally specific, it is crucial to understand what psychological and social factors predict involvement in cyberbullying either as a perpetrator, victim, or perpetrator/victim. The current study investigated whether problematic social media use and psychosocial well-being would predict cyberbullying and victimization among Mongolian adolescents.

Problematic social media use (PSMU) has been defined as a type of behavioral difficulty characterized by excessive use of social media, whereby individuals may experience a constant and impulsive urge to engage with social media, which is often to the detriment of their in-person relationships (Griffiths et al., 2014). In recent years, PSMU has consistently been found to be one of the significant predictors of both cyberbullying and victimization (Craig et al., 2020; Giordano et al., 2021; Lee et al., 2018). For example, a cross-national study involving 42 Western countries by Craig et al. (2020) reported that PSMU was the strongest predictor of both cyber-perpetration and victimization in most countries. Similarly, a systematic review (Lee et al., 2018) and a cross-sectional study from Spain (Feijóo et al., 2021) found that those who use social media intensively (more than 5 h per day) were more likely to be involved in cyberbullying experiences as either the bully, the victim, or both. This again shows that problematic social media use is a risk factor for cyber-perpetration, victimization, and perpetration/victimization. A recent cross-sectional study from North America also reported that social media addiction and being male were significantly related to increased risk of cyber-perpetration (Giordano et al., 2021). Although these studies used different terminologies such as intense social media use or social media addiction, the results indeed support that an increased amount of time on social media with some compulsive tendencies is associated with cyberbullying involvement.

The specific mechanism of how PSMU precedes cyber-perpetration has mainly been explained by the media exposure theory (Brown & Bobkowski, 2011; Craig et al., 2020). Media exposure theory posits that excessive use of social media results in increased exposure to violent content and aggressive self-expression thus normalizing and reinforcing aggression and bullying in the online context. In addition, previous research suggested that PSMU decreases the time for in-person extracurricular activities that develop psychosocial skills and moral senses which, in turn, help prevent risky and problematic behaviors such as cyberbullying (Jiang & Peterson, 2012). For cyber-victimization, deviant place theory seems to offer the best explanation that the increased time spent in the “dangerous” online environment increases the chances to get bullied (Stark, 1987). Thus, focusing on reducing the PSMU might help prevent cyberbullying incidents (e.g., Craig et al., 2020).

Moving further, taking the psychosocial factors that precede PSMU into consideration might help to tackle PSMU and eventually cyberbullying. Previous research suggests that adolescents with psychosocial distress are more prone to develop social media compulsion (Caplan, 2010; Faltýnková et al., 2020; Sela et al., 2020). Besides, social media has become an inseparable part of adolescents’ lives around the world. Therefore, solely focusing on online behaviors and hours spent online might be ineffective, especially during recent pandemic times.

According to the theory of problematic internet use and psychosocial well-being (Caplan, 2003, 2010), psychosocial well-being—perceived personal and social life quality (Negrini et al., 2014)—might determine whether to develop PSMU. The theory argues that adolescents with psychosocial distress are likely to have an increased preference for social media interactions as it provides them opportunities to express themselves anonymously and safely and that may facilitate the compulsion towards social media over time. Recent findings support a negative relationship between psychosocial well-being and PSMU, especially in adolescence (e.g., Boer et al., 2020; Heffer et al., 2019; Twenge, 2019). For example, Heffer et al. (2019) investigated adolescents’ and college students’ daily social media use and depressive symptoms over two years. Importantly, Heffer et al. tested both directional effects of the relationship (from depression to later frequent social media use and from social media use to later depressive score) to address the debate regarding whether psychosocial distress precedes PSMU or vice versa (see Keles et al., 2020; Orben, 2020). The study did not find any support for frequent social media use predicting depression for both adolescents and college students. Instead, depression predicted intensive social media use among adolescent girls. Another longitudinal study reported the same result that depressive symptoms during adolescence preceded increased preference for social media engagement after a year (Gamez-Guadix, 2014). Although these studies exclusively focused on time spent online, which technically is a part of PSMU symptoms, the longitudinal designs allowed the causal explanation that lower psychological well-being produces social media compulsion.

Furthermore, a recent study suggested that PSMU might result from decreased social support in in-person relationships and increased social support in online communications (Meshi & Ellithorpe, 2021). The researchers investigated the PSMU of American college students in relation to their perceived social support from both in-person and social media connections. Results revealed a negative association between PSMU and in-person social support and a positive association between PSMU and social media social support. Due to the cross-sectional design, a causal explanation cannot be drawn, as spending extensive time online might have reduced the opportunity for quality in-person relationships, or the low-quality in-person relationships might have triggered increased/problematic social media use. However, the latter explanation was supported by an extensive amount of qualitative, content-analysis research that individuals with various psychosocial distress such as mental illnesses (Naslund et al., 2014), suicidal ideation (Lavis & Winter, 2020), and loneliness (Pittman & Reich, 2016) tend to use social media for receiving and providing social support through self-disclosing posts and comments. These findings support that in addition to psychological distress such as depression (Gamez-Guadix, 2014), social distress leads to PSMU, possibly through reinforcing effects of perceived social support on social media.

However, the above arguments do not mean PSMU does not have consequences for adverse psychosocial outcomes. Increased time spent online has been found to produce fatigue (Dhir et al., 2018), sleep deprivation (Woods & Scott, 2016), and lack of beneficial time usage (Twenge, 2019). Twenge and Campbell (2019) reported that psychological well-being was significantly higher in adolescents who were classified as light users (i.e., less than one hour of use per day) when compared to individuals exhibiting heavy use (i.e., over 5 h per day). Moreover, Hunt et al. (2018) found that college students who restricted their social media use for over four weeks demonstrated significantly higher psychosocial well-being than the control group. That is to say, the relationship between psychosocial well-being and PSMU is undoubtedly complex, and it is important to acknowledge a possible dual model of adolescent psychosocial well-being that recognizes that positive and negative well-being are necessarily opposite ends of the same continuum (Suldo & Shaffer, 2008). However, lower psychosocial well-being seems to precede social media compulsion/PSMU due to the various possible reasons mentioned above.

Fig. 1
figure 1

Graphical illustration of the results of the mediation analysis for cyber-perpetration

Full size image

Taken as a whole, previous studies suggest that adolescents with psychosocial distress may spend increased time on social media with some compulsive tendencies (PSMU), which is likely to have consequences for involvement in cyberbullying. To date, only a limited number of studies tested whether PSMU mediates the relation between psychosocial well-being and cyberbullying involvement (Brighi et al., 2019; Kircaburun et al., 2019; Zsila et al., 2018). Particularly, based on the problem behavior theory (Jessor, 1987), Kircaburun et al. (2019) hypothesized that adolescents’ problematic behaviors such as PSMU and cyber-perpetration would be understood through their psychological health and social relations. Their results showed that depression directly predicted PSMU, which further predicted cyber-perpetration, while belongingness directly predicted both PSMU and cyber-perpetration. Cyber-perpetration was also related to being male and being younger. In short, the relationship between psychosocial distress, PSMU, and cyber-perpetration was stronger among high-school boys than the college ones. Somewhat similar results were found with Italian adolescents as emotional problems and low parental monitoring directly predicted cyber-perpetration and problematic internet use and indirectly predicted them through increased time spent online (Brighi et al., 2019). Therefore, the researchers concluded that lower psychological well-being was an underlying risk factor for cyber-perpetration, and parents may play a vital role through internet usage monitoring.

Lastly, while the above studies primarily focused on how psychosocial well-being impacts on cyber-perpetration through problematic social media or internet use, Zsila et al. (2018) investigated the relationship between problematic internet use and cyber-victimization and how perceived social support may have a protective effect. As hypothesized, problematic internet use predicted cyber-victimization, although with a small proportion of explained variance, and perceived social support, especially from parents, had a significant buffering effect. Consistent with findings from the wider cyberbullying literature (see review by Camerini et al., 2020), no significant age and gender differences in cyber-victimization were reported.

Fig. 2
figure 2

Graphical illustration of the results of the mediation analysis for cyber-victimization

Full size image

In summary, previous studies support that psychosocial well-being might predict adolescents’ cyberbullying involvement, and the social media use may mediate the relationship. However, previous studies tended to either focus on cyber-perpetration (e.g., Kircaburun et al., 2019) or cyber-victimization (Zsila et al., 2018). There is no study available that focused on cyber-perpetration and victimization simultaneously in relation to psychosocial well-being and social media use. Furthermore, previous literature, as highlighted in the present review, has tended to focus on these issues from a largely Western perspective. The complex social processes that underpin cyberbullying involvement are, as indicated in offline bullying research, likely to be impacted by social and cultural norms (Thornberg, 2015). Therefore, the current study will address these issues by considering cyberbullying phenomenon in the Mongolian context.

Present Study

Although adolescents’ cyberbullying has been considered a significant public health issue in Mongolia, where one in six Mongolians are of adolescent age 10 to 19 years (Communications and Regulatory Commissions, 2021), no published study is available on this specific population. As such there is a timely need to investigate the prevalence and experiences related to cyberbullying among Mongolian adolescents. The present study therefore addressed this gap by investigating cyberbullying involvement among Mongolian adolescents aged 14 to 18. Specifically, it considered the following research question (RQ): what are there associations between Mongolian adolescents’ psychosocial well-being, social media use, and cyberbullying involvement (i.e., perpetration and victimization)?

Our conceptualization of well-being stems from the perspective that well-being is broader than purely emotional health and should encapsulate both emotional and social constructs (Naci & Ioannidis, 2015). Therefore, in the present study, psychosocial well-being refers to the combination of (a) general psychological well-being (including states such as anxiety, depression, and perceived happiness) and (b) social well-being which exclusively includes the perceived social support from significant others as previous studies (e.g., Meshi & Ellithorpe, 2021) reported that perceived social support influences developing PSMU.

Cyberbullying involvement is complex (Lozano-Blasco et al., 2020), and risk factors for perpetration and victimization are somewhat similar (Craig et al., 2020; Feijóo et al., 2021), previous studies have tended to separately investigate cyber-perpetration (e.g., Kircaburun et al., 2019), or victimization (e.g., Zsila et al., 2018). Although there are clearly established links between lower psychosocial well-being and PSMU (Boer et al., 2020; Caplan, 2010; Heffer et al., 2019), as well as PSMU and cyberbullying involvement (Craig et al., 2020; Giordano et al., 2021; Lee et al., 2018), only a few studies investigated the relationships among all three variables simultaneously (e.g., Kircaburun et al., 2019). Specifically, previous studies seem to have been overly focused on PSMU in fighting cyberbullying, as it has been repeatedly found to be a strong predictor, through methods such as reducing time spent on social media using parental monitoring (e.g., Brighi et al., 2019). The rationale of those studies was that social media overuse causes problematic behaviors such as PSMU and cyberbullying. However, some studies showed that not all adolescents who are heavy users of social media (more than 5 h per day) develop PSMU or are involved in cyberbullying (Feijóo et al., 2021), and parental monitoring can result in even increased cyberbullying tendencies (Meter & Bauman, 2018). According to the theory of problematic internet use and psychosocial well-being (Caplan, 2010), it is possible that low psychosocial well-being might be responsible for PSMU, and that might, in turn, consequence cyberbullying involvement. In short, targeting PSMU in reducing cyberbullying involvement seems effective; however, including the psychosocial factors that trigger PSMU might produce a more comprehensive model. Based on the aforementioned literature, we therefore hypothesize that in the Mongolian context:

  • H1: PSMU will significantly mediate the relationship between psychosocial well-being and cyber-perpetration.

  • H2: PSMU will significantly mediate the relationship between psychosocial well-being and cyber-victimization.

Previous research into cyberbullying involvement outside of the Mongolian context has indicated that while age and gender may influence the likelihood of higher rates of perpetration (i.e., being young and male), significant effects in terms of victimization have not been apparent (Camerini et al., 2020). Given the paucity of cyberbullying data concerning Mongolian adolescents, the role of age and gender has not been afforded sufficient attention. Given that gender effects, in terms of higher rates of male perpetration, have been previously demonstrated in the Mongolian context (Optimal NMAX LLC, 2022), it is possible that age and gender effects may follow similar patterns to those demonstrated in non-Mongolian contexts. We therefore hypothesize for the Mongolian context:

  • H3: Being male and younger will relate to significantly more cyber-perpetration.

  • H4: Age and gender will not significantly relate to cyber-victimization.

Methods

Participants

Mongolian adolescents aged 14 to 18 who use at least one type of social media were invited to participate in this study. The invitation was distributed with an anonymous online link for the study materials, including the information sheet, informed consent document for participants and their parents, the questionnaire (hosted on Qualtrics), and the debrief, through a well-known website for Mongolian adolescents (www.Yolo.mn) for around two weeks during April 2021. We acknowledge that data collected via means of anonymous web-based participant recruitment can be prone to bots (Pozzar et al., 2020). Concerns about potential threats to study validity were mitigated by the use of Qualtrics which provides in-built mechanisms to prevent multiple responses from the same user. Institutional ethics practices prevented the collection and scrutiny of respondent IP addresses. Further checks for duplication of responses and data integrity were conducted during the process of data cleaning.

To indicate that consent was taken, all the participants and parents were asked to tick boxes on the informed consent page prior to the survey. Responses were collected from 782 participants; however, 106 participants were excluded as their responses only contained consent and demographic information. The final sample therefore consisted of a total of 676 adolescents, yielding a 86.4% response rate. Ninety-one percent were female (612 female and 64 male), 24.8% were from the countryside, and the remaining 75.2% were from the capital city Ulaanbaatar, with a relatively evenly distributed age range. Favorable ethical review was received from the School of Social Science Research Ethics Committee before the data collection procedure.

Measures

The original questionnaires were in English; thus, two-way translation was conducted, from English to Mongolian by the lead researcher and Mongolian to English by another native Mongolian speaker who was proficient in English to check the consistency of the meaning.

Psychosocial Well-being

Two questionnaires assessed adolescents’ psychosocial well-being. First, the Psychological General Well-being Index—a short version (Grossi & Compare, 2014)—was used to assess psychological well-being. The scale has 6 items, such as I was emotionally stable and sure of myself during the past month and I felt cheerful, lighthearted during the past month, that measures anxiety, depression, positive well-being, self-control, general health, and vitality. Responses range from none of the time (0) to all of the time (5). Higher scores indicate higher psychological well-being. The internal consistency coefficient of the original scale ranged from α = .80 to .94 (Grossi & Compare, 2014). In the current study, the internal consistency was α = .84 or highly consistent. Confirmatory factor analysis replicated a single-factor structure which demonstrated many of the requirements of an acceptable model. The comparative fit index (CFI) = .947 and Tucker-Lewis Index (TFI) = .876 exceed and were close to the acceptable value of .90 (Bryant & Yarnold, 1995), respectively. The root mean square error of approximation (RMSEA) = .109, which exceeds the recommended < .08 (Schubert et al., 2017), failed to meet the level to be acceptable. The significant chi-square, χ2 (9) = 81.58, p < .001, indicates some limitations in the fit of the data but is common when sample sizes exceed 200 (Schumacker & Lomax, 1996). All of the items loaded above the recommended .60 (Netemeyer et al., 2003) ≥ .62 and ≤ .75.

Second, 11 items from the Multidimensional Scale of Perceived Social Support (Zimet et al., 1988)—which assesses perceived social support from family, friends, and significant others—were used to measure social well-being. The scale includes items such as There is a special person who is around when I am in need, I get the emotional help and support I need from my family, and My friends really try to help me, and participants responded using a 7-point Likert scale: very strongly disagree (1) to very strongly agree (7). Higher scores indicate higher perceived social support. The test–retest reliability coefficient of the original 12-item scale was .85, and the internal consistency was α = .77 to .92 (Zimet et al., 1988). The internal consistency coefficient of this study was α = .89, indicating high consistency. Confirmatory factor analysis replicated the three-factor structure reported by Zimet et al. and represented an acceptable fit, CFI = .963, TLI = .940, RMSEA = .078, and items loaded ≥ .59 and ≤ .89, χ2 (41) = 208.41, p < .001.

A total score of the above two measures was used to indicate psychosocial well-being. The internal consistency for the combined measure was α = .89 or highly consistent. Confirmatory factor analysis including one factor for the Psychological General Well-being Index and three factors reflecting the structure of the Multidimensional Scale of Perceived Social Support indicated a good fit of the data CFI = .948, TLI = .930, RMSEA = .064 and items loaded ≥ .59 and ≤ .89, χ2 (113) = 429.43, p < .001.

Problematic Social Media Use

The Social Media Disorder Scale-9 (SMDS-9; Van den Eijnden et al., 2016), a 9-item, dichotomous response yes (1) and no (0) scale, assessed the PSMU. SMDS-9 was developed based on DSM-5 criteria for internet gaming disorder. Each item assesses a particular criterion, such as During the past years have you tried to spend less time on social media, but failed? assesses persistency of the social media compulsion and During the past years have you regularly found that you can’t think of anything else but the moment that you will be able to use social media again? assesses the preoccupation with social media. The data were analyzed continuously; thus, higher scores indicate higher PSMU. Confirmatory factor analysis suggested that the one-factor structure reported by Van den Eijnden et al. (2016) was not appropriate for the current data, CFI = .82, TLI = .715, RMSEA = .083 with items loading below the recommended value of .60, ≥ .39 and ≤ .52, χ2 (27) = 153.18, p < .001. Therefore, following Fung (2019), a two-factor structure was explored (factor one: items 1 (preoccupation), 2 (tolerance), 3 (withdrawal), 4 (persistence), 5 (displacement), and 8 (escape) considering the influence of an individual’s social media use on the self and factor two: items 6 (problem), 7 (deception), and 9 (conflict) considering the influence of social media use on an individual’s interactions with others) which yielded an improved fit, CFI = .93, TLI = .88, RMSEA = .046 with items loading on the corresponding factor ≥ .37 and ≤ .61, χ2 (26) = 76.85, p < .001. Fung (2019) suggests that such multidimensionality could be due to the influence of East Asian social contexts. Given that both factors are collectively indicative of established symptoms associated with behavioral addiction (Šablatúrová et al., 2022), the present study utilizes a single score in the main analysis in line with the approach adopted by recent East Asian research (Yu & Luo, 2021). The internal consistency coefficient of the original 9-item scale was α = .82, and this study’s was α = .71; therefore, it is considered reliable enough to conduct further analyses.

Cyberbullying Involvement

The Cyberbullying and Victimization Scale (CVS; Festl et al., 2015) assessed cyberbullying involvement. The scale assesses the frequency of exposure to cyber-perpetration and victimization in the last 12 months, through two sub-scales, a total of 11 items such as How often during the past year have you sent an insulting message to someone? for perpetration and How often during the past year did someone intentionally post embarrassing videos/photos of you on the internet? for victimization experiences. Answers are rated on a frequency basis: never (0), once (1), occasionally (2), and often (3). Higher scores on both subscales indicate a higher frequency of cyberbullying involvement. Participants can be classified as a perpetrator if answered “occasionally” or “often” at least once on the cyber-perpetration subscale, and the same logic applies to the cyber-victimization (Festl et al., 2015). These criteria classified the participants into four categories to explore the prevalence according to participant role (cyber-perpetrator, cyber-victim, both, and not involved). The internal consistency of the cyber-perpetration subscale was α = .64 or relatively lower. However, inter-item correlation ranged .21, or in the optimal range recommended by Briggs and Cheek (1986), and thus can be used for the data analysis. As opposed to that, the internal consistency of the cyber-victimization scale was acceptable as α = .75. Confirmatory factor analysis suggested a two-factor structure (factor one: cyber-perpetration items and factor two: cyber-victimization items) with fit statistics approaching acceptability, CFI = .89, TLI = .837, RMSEA = .069. However, several items loaded below .60 with item loadings between ≥ .38 and ≤ .68, χ2 (43) = 179.25, p < .001.

Data Analysis

Data analysis was conducted using SPSS.25 (IBM Corp, 2017). Firstly, some preliminary analyses were carried out for the sample characteristics, correlations among variables, and the cyberbullying involvement prevalence. Missing data accounted for less than 5% of the total dataset. An analysis of missing data was performed using Little’s MCAR Test (χ2 = 39.30 (47), p = 0.78). The non-significant result indicated the data were MCAR (Missing Completely at Random) which would allow for the use of listwise deletion rather than exclusion from the dataset (Allison, 2002). After that, the main analyses were conducted. Chi-square tests checked the gender and age differences in cyberbullying involvement according to participant role, identified following the procedure of Festl et al. (2015), and tested H3 and H4. For H1 and H2, the total scores for cyber-perpetration and cyber-victimization were used. The mediational analysis was performed using PROCESS Macro Model 4 for SPSS (Hayes, 2013), with a 95% confidence interval and 5000 bootstraps. If confidence intervals from the 5000 bootstrapped samples do not include zero, the direct and indirect effects are considered significant. Bootstrapping is considered to provide a robust and computationally intensive method of statistical analysis that overcomes violations in data assumptions (Özdil & Kutlu, 2019).

Results

Preliminary Analyses

Sample Characteristics and Correlations Among Variables

Ninety-one percent of the participants were female and the average age was 16 years (SD = .137, ranged from 14 to 18 years old). Significant correlations were found among the main variables (see Table 1). Specifically, psychosocial well-being was negatively associated with PSMU, r = −.267, p < .01; cyber-perpetration, r = −.160, p < .01; and cyber-victimization, r = −.266, p < .01, indicating that lower levels of psychosocial well-being were associated with higher levels of PSMU and higher levels of cyber-perpetration and victimization. PSMU was positively associated with both cyber-perpetration, r = .265, p < .01, and cyber-victimization, r = .235, p < .01, suggesting that higher levels of PSMU were associated with higher levels of cyber-perpetration and victimization. Cyber-perpetration was positively associated with cyber-victimization, r = .416, p < .01, in that higher levels of cyber-perpetration were correlated with higher levels of cyber-victimization. However, according to Akoglu (2018), the strength of the correlations was weak, except for the moderate correlation between cyber-perpetration and victimization.

Table 1 Descriptive statistics and correlations among variables (n ≥ 620 ≤ 676)
Full size table

Prevalence of Cyberbullying Involvement

Regarding cyberbullying involvement, participants were classified into four categories based on Festl et al.’s (2015) classification criteria. As a result, 6.7% (n = 45) were classified as pure cyber-perpetrators, 30.2% (n = 202) were pure cyber-victims, and 19.0% (n = 127) were both cyber-perpetrators and victims. Lastly, 43.5% (n = 294) were not involved in cyberbullying in any way. That is to say, for those who were involved in cyberbullying, pure cyber-victims were most common, followed by the dual-role, and being a pure cyber-perpetrator was the least common.

Main Analyses

Gender and Age Differences in Cyber-Perpetration, Victimization, and Perpetration/Victimization

In terms of pure cyber-perpetration, as hypothesized, 15.6% (n = 10) of males and 5.7% (n = 35) of females cyber-perpetrated others; and this gender difference was statistically significant, X2(1, n = 668) = 7.67, p = .006, phi = −.12. The most common behavior of cyberbullying for both females and males was to send an offending message on social media in which males (M = 1.80, SD = .979) tended to send offending messages significantly more often than females (M = 1.49, SD = .804), t(72.3) = −2.40, p = .02, g = .37. Other ways of cyberbullying, such as spreading rumors or sending embarrassing pictures/photos of someone, were less common and no gender difference was found (p > .05). Regarding pure cyber-victimization (H4), 26.5% (n = 17) of males and 30.2% (n = 185) of females were cyberbullied and the gender difference was not significant, X2(1, n = 668) = .217 p = .64, phi = .02. Lastly, 26.5% (n = 17) of males and 18.2% (n = 110) of females were found to be experienced both perpetration and victimization, and the gender difference was not significant, X2(1, n = 668) = .210, p = .15, phi = −.06. That is, while males and females were equally likely to experience pure cyber-victimization and perpetration/victimization on social media, males were significantly more likely to purely cyber-perpetrate others through sending offending messages. In terms of age differences (H3 and H4), there were no significant age differences in cyber-perpetration, X2(4, n = 668) = 5.43, p = .25, phi = .09; cyber-victimization, X2(4, n = 668) = .49, p = .97, phi = .03; and perpetration/victimization X2(4, n = 668) = 4.27, p = .37, phi = .08. Thus, H4 was supported and H3 was partially supported.

Mediation Analyses

The PROCESS Macro program for SPSS developed by Hayes (2013) was used to test the mediational hypotheses. Firstly, we ran the analysis using the total scale score for cyber-perpetration as the outcome variable, with psychosocial well-being as the predictor and PSMU as the mediator variable. After that, the same analysis was run for cyber-victimization, using the total scale score for cyber-victimization as the outcome variable.

Table 2 outlines the total, direct, and indirect effects exploring PSMU as a mediator in the relationship between psychosocial well-being and cyber-perpetration. There was a significant negative indirect effect supporting H1 (β = −.08, BCa CI [−.012, −.004]), although this effect was only partial. Psychosocial well-being negatively predicted PSMU (β = −.035) such that lower levels of psychosocial well-being predicted higher levels of PSMU which in turn predicted higher levels of cyber-perpetration (β = .220) (Fig. 1).

Table 2 Total, direct, and indirect effects (mediation analysis for cyber-perpetration)
Full size table

Table 3 outlines the total, direct, and indirect effects exploring PSMU as a mediator in the relationship between psychosocial well-being and cyber-victimization. There was a significant negative indirect effect supporting H2 (β = −.08, BCa CI [−.014, −.004]), although this effect was only partial. Psychosocial well-being negatively predicted PSMU (β = −.035) such that lower levels of psychosocial well-being predicted higher levels of PSMU which in turn predicted higher levels of cyber-victimization (β = .237) (Fig. 2).

Table 3 Total, direct, and indirect effects (mediation analysis for cyber-victimization)
Full size table

Discussion

The present study was one of the first to investigate cyberbullying involvement among Mongolian adolescents. Over half (56.5%) of the current sample was involved in cyberbullying either as a victim, bully, or both. In which, being pure cyber-victim was most common (30.2%), followed by bully/victim (19.0%), and being pure cyber-bully was least common (6.7%). These results replicated, in a different country, previous findings that cyber-victims are more common than cyberbullies (Guo et al., 2021; Wang et al., 2009), suggesting that bullies are likely to have more than one victim. It is likely, however, that the bullying categories captured here may be more socially complex than the quantitative data permits us to understand (Thornberg, 2015). Indeed, the high number of dual roles in the current sample and the significant correlation between cyber-perpetration and cyber-victimization indicate some interrelatedness of being both victim and bully for adolescents in Mongolia. Previous research explained why victims later become bullies regarding the normalization effect of repeated exposure to bullying (Guo et al., 2021; Lozano-Blasco et al., 2020) and retaliation motives (Runions et al., 2018). Furthermore, qualitative research implies that victims may adopt bullying-type behaviors as they strive to gain recognition and a sense of belonging, with the display of ‘deviant’ behaviors acting as a means to break out of the social confines of victimhood (Thornberg, 2011).

In the present study, males tended to cyberbully others more than females, in line with previous findings from other countries (Giordano et al., 2021; Kircaburun et al., 2019) and from Mongolia (Optimal NMAX LLC, 2022). No gender difference was found for cyber-victims, as hypothesized. The most common behavior of cyberbullying was sending an offending message on social media. It might be that harassing or discouraging interactions have become a social media norm among Mongolian adolescents, especially among males. The adolescents also might be unaware that such behavior is an unacceptable element of bullying. Moreover, there could be some male masculinity-related Mongolian cultural influences (Hurst, 2001; Rarick et al., 2014). Traditionally, aggressive masculine behaviors are somewhat acceptable for Mongolian males, but not for females. Due to that, abusive and violent behaviors (towards both males and females) tended to be common among Mongolian males (Hurst, 2001). Our findings seem to suggest that this gendered cultural effect might be still present in Mongolia to some extent. Further studies are encouraged to investigate these hypotheses. Nevertheless, anti-cyberbullying strategies in Mongolia may need to focus more on males and educate that sending offending messages is bullying behavior and harmful to both their own and the victim’s well-being (Cañas et al., 2020; Gámez-Guadix et al., 2013). Additionally, contrary to the previous findings (Giordano et al., 2021; Kircaburun et al., 2019), no age differences were found for cyber-perpetration and victimization. While this may suggest that anti-cyberbullying programs in Mongolia could be appropriate to all ages, we acknowledge that the relatively narrow adolescent age range of the sample may not fully reflect the individual differences that might influence the construction of such interventions for young people. We would also suggest that future research extends this line of inquiry by involving young people as experienced peer-researchers with a unique role as ‘insiders’ who could shape research designs and explanations (Dunn, 2015).

Furthermore, the main hypotheses that PSMU mediates the relationship between psychosocial well-being and cyberbullying involvement were supported. The mediations were partial for both cyber-perpetration and victimization models, in line with Kircaburun et al. (2019). While Kircaburun et al. solely measured cyber-perpetration as the outcome variable, the current study showed lower levels of psychosocial well-being and higher levels of PSMU could lead to both cyber-perpetration and cyber-victimization. However, the theoretical explanations of how psychosocial distress and PSMU lead to cyber-perpetration and victimization could be different, as suggested by Craig et al. (2020). Particularly, for cyber-perpetration, the problem behavior theory (Jessor, 1987) seems to offer the best explanation. It argues that adolescents engaging in problematic behaviors, such as bullying and behavioral addiction, usually do so because of the decreased psychological and environmental well-being. In addition to that, PSMU may result in cyber-perpetration as constant exposure to aggressive content and interactions on social media normalizes and reinforces aggression or bullying, as stated in the exposure theory (Brown & Bobkowski, 2011). PSMU also may result in decreased opportunity for in-person extracurricular activities that help develop moral senses and social skills, thus increasing the risk of involvement in problematic behavior such as cyberbullying (Jiang & Peterson, 2012).

The deviant place theory (Stark, 1987) offers an explanation of how PSMU results in cyber-victimization. The theory suggests that frequent exposures to dangerous areas increase the risk of becoming a victim. To put it in another way, adolescents with lower levels of psychosocial well-being might use social media more compulsively to overcome the distress as stated in the theory of psychosocial well-being and problematic internet use (Caplan, 2003, 2010), and the compulsive use of social media might increase the chances of becoming a cyberbullying victim. In summary, the results seem to support the hypothesis that adolescents with psychosocial distress use social media more compulsively, increasing the chance and tendency to be involved in cyberbullying either as a victim or bully. However, it should be noted that, in line with Kircaburun et al. (2019), the effect sizes of psychosocial well-being and PSMU were small for both cyber-perpetration and cyber-victimization; thus, other factors that are not included in the current study could be more stronger predictors.

Limitations

This study is not without limitations. First, the cross-sectional design limits the causal inferences that can be made. For example, psychosocial well-being was more strongly negatively related with cyber-victimization than cyber-perpetration, and the direction of this relationship was unknown. It could be that instead of lower levels of psychosocial well-being leading to cyber-victimization through PSMU as hypothesized in this study, cyber-victimization experiences could be leading to lower well-being as suggested in some studies (e.g., Feinstein et al., 2014). Future studies are encouraged to test the assumptions and inferences of this study through longitudinal methods. Second, the majority of the participants (91%) were females while 50.8% of the total populations are females (Kemp, 2022). Data collection over social media might be the reason, as females tend to be more active on social media than males (Leonhardt & Overa, 2021). Nevertheless, future studies are recommended to demonstrate the gender ratio that can represent the specific population. Third, this research was carried out during the COVID-19 pandemic. Thus, adolescents may have been using social media more compulsively (Deslandes & Coutinho, 2020) and experiencing more psychosocial distress because of the pandemic and the lockdowns (Magson et al., 2021). This context-related effect was not controlled as there was no comparable data from this specific population. Therefore, after-pandemic research is needed. Fourth, the research did not explore the role of bystanders. Previous research has highlighted that bystanders, according to their responses to an episode of cyberbullying, can exacerbate the impact of experiencing cyber-victimization (DeSmet et al., 2019). Therefore, future research should consider the role of bystanders. Fifth, while we took steps to avoid the infiltration of bots through the prevention of multiple responses and data scrutiny, recent advances in survey technology may provide more rigorous options such as Google reCAPTCHA scores. Furthermore, future studies may wish to consider using mischievous responder measures and analysis (Robinson-Cimpian, 2014). Sixth, some of the indicators of fit for the confirmatory factor analyses were on the lower end of acceptable, particularly for the measurement of involvement in cyberbullying. However, previous research has acknowledged a lack of exploration of the psychometric properties within cyberbullying research (Thomas et al., 2015). Consequently, researchers need to be mindful of this in future studies. Finally, while the use of a total score for PSMU was in line with previous discussions in the literature (Yu & Luo, 2021), future research should be mindful of the need to explore the two-factor structure indicated in the CFA to determine the potential for alternative interpretations of the role of PSMU.

Future Research Directions and Implications for Practice

As this is one of the first studies to explore cyberbullying in Mongolian adolescents, it highlights that cyberbullying is an issue and suggests more research is needed in Mongolia to explore the social and cultural complexities of why adolescents engage in cyberbullying (e.g., with what motives) and how they define and perceive cyberbullying roles and behaviors in the Mongolian cultural context. For this, we suggest that researchers look towards qualitative approaches to research that acknowledge the relevant peer culture (c.f. Thornberg, 2011). Moreover, the findings show that the same risk factors can contribute to cyber-perpetration and victimization. Specifically, targeting PSMU could effectively deal with cyber-perpetration and victimization in line with previous findings (e.g., Craig et al., 2020); however, psychosocial well-being might be contributing to both PSMU and cyberbullying involvement. Therefore, for the policymakers, school administrators, and other stakeholders in Mongolia, focusing on promoting adolescents’ psychosocial well-being and healthy social media use may be an effective approach to prevent and deal with cyber-perpetration and victimization. To this end, specific and appropriately adapted programs on promoting well-being and healthy social media use (e.g., the development of educational frameworks for adolescent online safety such as Project Evolve (2023)) are recommended for a Mongolian context. Furthermore, the findings of this study could be used to develop and inform anti-cyberbullying awareness-raising in Mongolia, for instance, through social media campaigns, drawing inspiration from successful initiatives developed in non-Mongolian contexts (e.g., campaigns by Ditch the Label from the UK). Successful and culturally specific anti-bullying interventions are particularly important because of the long-term consequences of involvement in bullying (Wolke & Leyera, 2015). However, we are also mindful that the majority of the research cited in this paper are from studies conducted in different cultural contexts to Mongolia and that, in some cases, these studies do not always consider the cultural context of bullying.

Conclusion

The present study explored cyberbullying involvement among Mongolian adolescents in relation to their psychosocial well-being and social media use. Findings suggest that cyberbullying involvement might be a common experience among this population as more than half of the participants were involved in cyberbullying either as a bully, victim, or bully/victim. Males tended to cyberbully more by sending offending messages. Findings also suggest that lower levels of psychosocial well-being predict higher levels of PSMU which further predicts cyber-perpetration and victimization. Therefore, it is recommended for the relevant stakeholders in Mongolia to focus on adolescents’ psychosocial well-being and social media use as well as to initiate anti-cyberbullying awareness-raising projects.