Social media threats and health among adolescents: evidence from the health behaviour in school-aged children study

The social media have brought about a new era of victimization, involving unique challenges and consequences. This is particularly evident in the phenomenon of cyberbullying, defined broadly as bullying via electronic means [26, 27]. Social media allow perpetrators to target their victims at any time and any place (either in front of large audiences or privately) via 24/7 accessibility [28]. Additionally, the possibility of remaining anonymous on these platforms may embolden perpetrators to continue bullying with no fear of repercussions [29]. The absence of face-to-face cues on social media hides the negative consequences of cyberbullying [30], and without this critical feedback, aggressive behaviour may be more likely to recur [12, 31]. In addition, continuous exposure to online aggression can lead individuals to view this behaviour as more acceptable through reinforcement and role modelling, especially if it is rewarded socially [12, 32]. According to a recent systematic review, the prevalence of cyberbullying victimization ranges between 14% and 58% among adolescents [33]. The authors note that one reason for the inconsistency in the prevalence rates is the differences in recall periods across studies (lifetime, last year, last month, etc.); hence, more nuanced research is needed to determine how often adolescents are exposed to cyberbullying [33].

Discrimination via social media, which includes online sexual harassment [9] and racial discrimination [13], presents another significant threat to adolescents. Previous studies have primarily defined online sexual harassment as unwanted sexual behaviour that occurs electronically, such as sending unsolicited sexual messages, images, or requests, or having sexual messages or images shared without permission [9, 34]. Online racial discrimination, on the other hand, refers to any unfair or prejudicial online act based on, for example, race, skin colour, or ethnicity [13]. Noting the easily accessible nature of social media (such that anyone can approach anyone), the rapid dissemination of information, and the broad audience reach, studies have suggested that social media have increased encounters with both forms of discrimination, beyond parental oversight [8]. Social media platforms may amplify discrimination by allowing perpetrators to target multiple victims simultaneously, while remaining anonymous and maintaining a physical distance [9, 35]. Like cyberbullying, the prevalence of online sexual harassment has been inconsistent across adolescent studies (1–59%), partly due to differences in recall periods [36]. The literature on racial discrimination, for its part, has mainly focused on vulnerable populations (e.g., adolescents of colour) [13] or adults [35]. Consequently, the prevalence of adolescents’ encounters with racial discrimination remains relatively unknown. One of the few studies on this issue found that out of adolescents aged 12–16, 17% reported at least monthly exposure to hate messages attacking certain groups or individuals [8].

One social media threat related to both cyberbullying and discrimination – as shown in previous studies [e.g., 37] – is unauthorized sharing of sensitive material (e.g., sexually explicit images). Such material can circulate through adolescent social networks and can be difficult to eradicate from the web [37]. Sensitive material may be shared willingly (e.g., in a relationship), and later disseminated nonconsensually as an act of revenge (e.g., during breakup) [34]. Furthermore, sensitive material can be acquired through phishing attempts by third parties [38]. In both instances, the sensitive material can be used to blackmail the victim [34, 39]. In 2020, 11% of adolescents aged 9–16 had experienced personal data misuse, such as somebody using personal information in a way the victim did not like, or somebody using a person’s password to access personal information over the past year [8]. Notably, 20% of adolescents did not know how to change their privacy settings [8], a factor that may place adolescents in a vulnerable situation in terms of privacy.

According to Southwell et al. [14], the current generation of adolescents faces a massive proliferation of rapidly spreading misinformation, i.e., inaccurate or misleading information running contrary to the best scientific evidence [14, 40]. The concern over misinformation was predominantly evident during the COVID-19 pandemic, when a wave of information on the COVID-19 virus spread, especially on social media [23]. Furthermore, a large-scale European study has indicated clear disparities in capabilities to access valid information, with 41% of adolescents reporting an inability to assess the validity of online information [8].

Concerns have also been raised about exposure to risky online content, such as the sale, distribution, and misuse of drugs [41‐43]. With increased societal restrictions and parental surveillance over access to most substances, the social media may provide a novel environment for the display of adolescents’ risk-taking proclivities [44, 45]. For example, initial evidence points toward the increased use of social media to buy and sell drugs; in this case, the media facilitate easy access to groups where products are distributed and allow contactless deliveries to end-consumers [46]. Social media also foster participation in risky social media challenges [43]. According to previous theories and models, such as the ‘super peer’ theory [47], and the Facebook Influence Model [41], the social media context amplifies peer influence, due to the increased volume of content portraying risky behaviours, and the quantifiable reinforcement of such behaviour in the forms of ‘likes’ and comments. The social media context may also exacerbate sensation-seeking by framing risky challenges as exciting and pleasurable [42]. In Europe, on average, 8–17% of adolescents aged 12–16 report that they have faced harmful content online at least monthly. Approximately 10% of adolescents mentioned having viewed content on experiences of taking drugs, how to commit suicide, or how to physically harm oneself [8].

Popular adolescent social media platforms contain an abundance of appearance-focused content [48]. This content tends to promote muscular and athletic ideals for males, such as large biceps, a V-shaped torso, and visible abs. For females, it endorses thin and curvaceous ideals, such as a lean physique, low body fat, and a thin waist with a prominent bottom or bosom [49]. One important way in which the social media differ from traditional media is that the content is user-generated [48]. Adolescents report spending considerable time and effort on their images to represent the ‘best’ version of themselves, a process which is enhanced by in-built photo-editing tools [50]. Adolescents are thus exposed to idealized and possibly edited content of people, including peers and influencers, through their use of social media [48]. Furthermore, 12% of people aged 12–16 had come upon content on ‘ways to be very thin’ at least monthly [8].

There is a growing consensus among researchers that adolescent exposure to social media threats is a complex phenomenon with several explanatory factors. Factors such as gender, age, and capabilities play a role in the threats to which adolescents are exposed. For example, girls are more likely to report online sexual harassment [10] and to encounter appearance-focused social media content [51, 52], whereas boys seem to be more likely to seek out violent material [8], and to have a lower perception of the risks related to publication of data and photographs [53]. Adolescent age has further been hypothesized as explaining social media threat exposure, on the grounds that periods of increased sensitivity to certain threats are likely to occur in relevant developmental windows [15, 54]. For example, Smahel et al. [8] found that adolescents aged 15–16 are more likely than those aged 9–11 to encounter threats such as cyberbullying and personal data misuse. In addition, disparities in capabilities to protect oneself from social media threats have been identified [8]. Emotional intelligence, defined as the ability to identify and comprehend one’s own emotions as well as the emotions of others, and to utilize this understanding to effectively regulate one’s own behaviour and relationships [55], has been suggested as a protective factor against social media threats [56, 57].

In addition, social factors such as family affluence, family support, and friend support have been linked to adolescent social media threat exposure. Studies on young people have found those with lower socioeconomic status to be more likely to report negative social media experiences, such as receiving hurtful messages, or having pictures/videos shared without their consent [58, 59]. Conversely, higher family support has been found to significantly reduce adolescents’ risk of encountering social media threats. For example, a review study by Elsaesser et al. [60] found that higher family support was consistently associated with lower cyberbullying victimization and perpetration. The relationship between friend support and social media threats, however, is expected to be double-edged. On the one hand, a higher level of friend support has been shown to protect young people from social media threats such as cyberbullying [61]. On the other hand, friend support can enhance the likelihood of encountering and participating in risky behaviours (e.g., dangerous social media challenges) [43].

Research also suggests that social media–related disorders contribute to exposure to social media threats. For example, problematic social media use (PSMU) – as indicated by addiction-like symptoms (i.e., withdrawal, conflict, preoccupation) – has been associated with cyberbullying [12], and exposure to misinformation/misconceptions on COVID-19 during the outbreak [62]. The idea of PSMU being associated with social media threats is supported by social theories such as the Problem Behaviour Theory, which suggests that certain risk behaviours are interconnected and contribute to vulnerability [63, 64]. Concerns have also been raised about adolescents who use social media to intensively communicate with strangers. Although uncommon, strangers may pursue ill intentions such as solicitation and harassment [65]. Adolescents may be unable to identify malicious intents due to their sensitivity to acceptance, feelings of rejection, and a lack of self-awareness [5].

Adolescents have experienced a significant increase in depression and anxiety over the last decade, and the social media have been suspected as a primary cause [66, 67]. However, the evidence is conflicting. On the one hand, numerous reviews have established a connection between social media use and negative health outcomes among young people [e.g.,68, 69]. On the other hand, a recent umbrella review concluded that the association between social media use and poor adolescent wellbeing was ‘weak’ and ‘inconsistent’ [70]. There have been calls for research to shed light on these inconsistent findings, focusing on the mechanisms that could make social media harmful to adolescents’ health [67, 70‐73]. Encounters with social media threats have been proposed as one such mechanism [74]. Thus far, studies among young people have found cyberbullying, sexual harassment victimization, and racial discrimination to be associated with negative health outcomes such as depressive symptoms and anxiety [9, 10, 75, 76]. Misinformation can negatively influence adolescents’ health and health behaviour by eroding their judgement, and by shaping the precursors of their intentions [14]. These can include their attitudes toward behaviour, for example, in terms of approving or disapproving [14]. It may further disrupt their feelings of security, as has happened, for instance, via content related to COVID-19 [77]. Hence, misinformation has been associated with negative moods, anxiety, and distress [77, 78]. Furthermore, threats such as alcohol-related content, and harmful social media challenges have been related to harmful behavioural choices [14, 41, 42]. Idealized appearance-focused content on the social media, for its part, provides adolescents with opportunities to internalize prescriptive ideals, self-objectify, and engage in negative upward appearance comparisons, which could trigger body dissatisfaction [48, 79]. All this would suggest that encounters with social media threats would be a stronger determinant of negative health among adolescents than social media use alone [73].

The current state of research leaves gaps in our understanding. We lack a comprehensive understanding of how frequently adolescents are exposed to various social media threats in Finland, or within the broader empirical context. There has been relatively little research on the prevalence of certain threats (e.g., dangerous social media challenges, or the sale or distribution of drugs), and the prevalence rates for certain threats (e.g., cyberbullying and sexual harassment) have been inconsistent across studies due to varying reporting frequencies [33, 36]. Some studies have explored the association between individual and social factors and social media threats under specific conditions; however, there has so far been no comprehensive examination of a broad set of individual and social factors in relation to various social media threats [6‐8]. Furthermore, despite a recent surge in studies on social media use and health, our understanding of the mechanisms through which social media use might harm adolescent mental health and wellbeing remains limited [67, 70‐73]. This emphasizes the need to determine how various social media threats are associated with health outcomes in adolescence. To address these research gaps, the present study aimed to evaluate adolescents’ encounters with nine social media threats at distinct intervals: ‘never’, ‘monthly’, ‘weekly’, and ‘daily’, and their association with individual and social factors, PSMU, online communication with strangers, and health outcomes. Thus, by utilizing a nationally representative sample of Finnish adolescents, the following research questions were addressed:

(RQ1) How prevalent are social media threats (cyberbullying, sexual harassment, racism, unauthorized distribution of sensitive material, phishing attempts, misinformation, sale or distribution of drugs, harmful or dangerous social media challenges, content causing appearance pressures)?

(RQ2) What are the associations between exposure to social media threats and (i) individual factors (gender, age, emotional intelligence), (ii) social factors (family affluence, family support, friend support), (iii) PSMU, (iv) online communication with strangers?

(RQ3) How are social media threats associated with health (self-rated health, depressive feelings, anxiety)?

Based on previous research, the following hypotheses were formed:

Nationally representative data were collected from Finnish adolescents in 2022 as part of the international Health Behaviour in School-aged Children Study (HBSC). The data were collected through anonymous voluntary standardized questionnaires administered to young people aged 11, 13, and 15 via school-based surveys. A stratified random cluster sampling design was used, and the data collection followed guidelines prescribed by the HBSC research protocol [80]. Ethical approval for the study procedures was obtained from the institutional ethics committee of the University of Jyväskylä.

In total, the sample comprised of 2288 Finnish boys (n = 1117; 48.8%) and girls (n = 1171; 51.2%) between the ages of 11 (n = 904; 39.5%), 13 (n = 764; 33.4%), and 15 (n = 620; 27.1%).

Missing data ranged between 1.4% (gender) and 15.6% (problematic social media use). To overcome the potential bias associated with listwise deletion, we utilized multiple imputation by chained equations. Multiple imputation reduces the potential bias related to missing data even when the percentage of missing data is high [101]. The missing data were imputed on the basis of available data on other included study variables. Five imputations were conducted, in line with the suggestions of Madley-Dowd et al. [101]; thus, all 2288 respondents were retained for the analyses.

The associations between individual and social factors, PSMU, online communication with strangers, and social media threats were tested using fixed effects multinomial logistic regression analyses, and reported as odds ratios (ORs). For the social media threats, ‘Never’ was used as the reference category. A separate analysis of 15-year-olds was performed for emotional intelligence, as the variable was only measured in this age group. Variables were added to the models hierarchically, and adjusted effects were reported.

Fixed effects binary logistic regression analyses were conducted to study the association between social media threats and health outcomes. The regression models were performed on each health outcome separately, and the analyses were adjusted for gender, age, and family affluence. All fixed effects logistic regression models were tested for the clustering effect of schools in the data. The analyses were conducted with IBM SPSS Statistics 28.0 [102].

As shown in Table 1, the two most prevalent social media threats encountered by adolescents daily were misinformation (12.9%) and content causing appearance pressures (9.1%). On a weekly basis, the most prevalent social media threats were misinformation (44.2%) and harmful or dangerous social media challenges (22.3%). In terms of monthly exposure, the most prevalent social media threats were unauthorized distribution of sensitive material (27.7%) and harmful or dangerous social media challenges (26.8%). The least prevalent social media threats (in terms of the ‘never encountered’ option) were cyberbullying (79.5%) and sexual harassment (77.7%).

As indicated by Table 2, significant associations were identified between social media threats and individual factors. In terms of encountering social media threats daily, seven out of the nine threats were more likely to be reported by boys, including cyberbullying (OR = 0.28, CI 95% = 0.15–0.53), sexual harassment (OR = 0.51, CI 95% = 0.27–0.96), racism (OR = 0.47, CI 95% = 0.31–0.70), unauthorized distribution of sensitive material (OR = 0.37, CI 95% = 0.24–0.57), phishing attempts (OR = 0.18, CI 95% = 0.11–0.32), misinformation (OR = 0.47, CI 95% = 0.34–0.66), and harmful or dangerous social media challenges (OR = 0.42, CI 95% = 0.27–0.65). By contrast, content causing appearance pressures (OR = 6.71, CI 95% = 4.51–9.98) was the only threat more likely to be reported by girls.

In terms of weekly exposure, cyberbullying (OR = 0.32, CI 95% = 0.21–0.48) and phishing attempts (OR = 0.43, CI 95% = 0.32–0.57) were more likely to be reported by boys, whereas exposure to racism (OR = 1.40, CI 95% = 1.10–1.79) and content causing appearance pressures (OR = 4.79, CI 95% = 3.65–6.29) were more likely to be reported by girls. Girls were also more likely to report monthly exposure to sexual harassment (OR = 2.53, CI 95% = 1.75–3.64), racism (OR = 1.89, CI 95% = 1.49–2.41), misinformation (OR = 1.34, CI 95% = 1.02–1.75), harmful and dangerous challenges (OR = 1.89, CI 95% = 1.53–2.35), and content causing appearance pressures (OR = 3.71, CI 95% = 2.84–4.84).

Adolescents aged 15 self-reported daily (ORs 2.82–20.89) and weekly (ORs 2.05–11.11) exposure to every social media threat more than did those aged 11 (Table 2). Similarly, compared to 11-year-olds, 13-year-olds were more likely to encounter six out of the nine social media threats daily (ORs 2.00–7.71), every social media threat weekly (ORs 2.20–5.21), and eight out of the nine threats monthly (ORs 1.42–3.12). Adolescents with higher emotional intelligence were less likely to report daily exposure to cyberbullying (OR = 0.40, CI 95% = 0.23–0.72), sexual harassment (OR = 0.34, CI 95% = 0.17–0.70), racism (OR = 0.57, CI 95% = 0.36–0.89), unauthorized distribution of sensitive material (OR = 0.57, CI 95% = 0.35–0.94), and phishing attempts (OR = 0.49, CI 95% = 029–0.84).

Adolescents with higher family affluence were more likely to report daily encounters with misinformation (OR 2.24, CI 95% = 1.25–4.03), sale or distribution of drugs (OR = 1.85, CI 95% = 1.03–3.35), and content causing appearance pressures (OR = 1.81, CI 95% = 1.00–3.29), weekly exposure to content causing appearance pressures (OR = 1.67, CI 95% = 1.05–2.65), monthly encounters with harmful social media challenges (OR = 1.76, CI 95% = 1.21–2.57), or content causing appearance pressures (OR = 2.63, CI 95% = 1.66–4.18; Table 3).

Adolescents with higher family support were less likely to report daily (ORs 0.60–0.78) and weekly (ORs 0.72–0.87) exposure to eight out of the nine social media threats, and monthly (ORs 0.78–0.86) exposure to three social media threats. In terms of social support from friends, adolescents with higher support were more likely to report daily encounters with the sale or distribution of drugs (OR = 1.19, CI 95% = 1.05–1.36), but less likely to report daily (OR = 0.78, CI 95% = 0.64–0.94) and weekly (OR = 0.75, CI 95% = 0.65–0.85) encounters with cyberbullying.

Adolescents with PSMU were more likely to report daily (ORs 3.00–5.66) and weekly (ORs 1.58–4.72) exposure to every social media threat except misinformation (Table 4). For example, problematic users were more than five times as likely to report daily exposure to cyberbullying (OR = 5.64, CI 95% = 2.97–10.69) and sexual harassment (OR = 5.66, CI 95% = 3.07–10.42). Those who reported intensive online communication with strangers were more likely to encounter eight out of the nine social media threats daily (ORs 2.03–6.02), as well as exposure to four social media threats weekly (ORs 1.96–2.91).

Adolescents who encountered any of the social media threats daily or weekly were more likely to report having poor self-rated health, frequent depressive feelings, and frequent anxiety symptoms, as compared to those who never reported such encounters. For instance, those exposed to misinformation daily were almost three times as likely to report poor self-rated health (OR = 2.83, CI 95% = 1.68–4.76), and approximately four times as likely to report frequent depressive feelings (OR = 4.15, CI 95% = 2.63–6.54) and frequent anxiety symptoms (OR = 3.78, CI 95% = 2.47–5.78; Table 5). Furthermore, adolescents who encountered any of the threats as infrequently as once a month were more likely to report having frequent depressive feelings than those who never experienced such threats (ORs 1.33–2.48). Similarly, adolescents with monthly exposure to eight out of the nine threats were more likely to report frequent anxiety symptoms (ORs 1.62–2.60). Adolescents exposed monthly to cyberbullying, sexual harassment, or phishing attempts were more likely to report poor self-rated health (ORs 1.54–1.97).

The present study had several strengths, including a large-scale nationally representative sample of adolescents and the use of validated instruments, plus a carefully considered distinction between different social media threats. Furthermore, to our knowledge, this is the first study to provide evidence on the association between PSMU and several other threats (e.g., harmful social media challenges), thus, opening up novel horizons for future studies and interventions. However, the findings should be interpreted with several limitations in mind. Firstly, the study’s cross-sectional design does not allow causal inferences. Secondly, it can be argued that self-reporting instruments may not give an objective view of adolescent exposure to social media threats, and that self-reporting measures of media experiences may not be a legitimate substitute for more objective measures [111]. For example, in terms of cyberbullying, the power imbalance between the perpetrator and the victim cannot be precisely measured via the type of self-reporting instrument used in this study. For instance, researchers argue that power might behave differently in the online context and that power dynamics can manifest through technological proficiency or possession of harmful content [112, 113]. Hence, it could be interpreted that any individual who can exploit technology to harm others holds a position of power, at least temporarily, in relation to the victim of the assault [113]. Similarly, self-reporting instruments may not provide an objective reflection of certain individual and social factors, given that not all individuals will necessarily perceive, for instance, emotional intelligence in the same way. However, the information given by self-reporting instruments is necessary if one is to explore individuals’ personal experiences and views [114] on their emotional intelligence. One must also bear in mind that experiences of social media threat exposure are individual and subjective (as in the case of cyberbullying); hence, they need to be investigated via measures considering individual experiences (as in this study). Nonetheless, the information could have been enriched by the views of multiple informants, including peers, parents, or teachers.

Finally, one must be cautious about generalizing the results beyond the study populations (e.g., to non-white, and low-income countries). To overcome these limitations, future studies should employ cross-national study settings, wider study populations, and longitudinal research. Furthermore, moderation and mediation approaches could be applied to better understand (i) the factors protecting against social media threats, and (ii) how social media threats operate in the association between the intensity of social media use and health, and in the associations between various social media activities and health outcomes. There is also a need to investigate how social media threats co-occur and interrelate, for example, whether being exposed to one social media threat increases the likelihood of being exposed to many. One could also seek to determine whether certain types of threats accumulate for specific individuals, and how the individuals themselves act or react (e.g., as regards cyberbullying perpetration, sexual harassment perpetration, and the sharing of misinformation) on social media. Person-oriented approaches such as Latent Class Analysis (LCA) would be advisable in this regard.

Our study indicates that intervention and health promotion efforts are needed to reduce adolescent exposure to social media threats and associated negative health outcomes. The efforts should consider the individual and social differences among adolescents (the aim being to promote equity by ensuring that adolescents in vulnerable situations benefit proportionately more from such efforts) [see 105]. The measures taken could aim to support resources such as emotional intelligence and family support against social media threats. Furthermore, we suggest that, in particular, governments and service providers should act and collaborate to reduce adolescent encounters with social media threats. The negative impacts of social media threats on health could be mitigated by directing resources to vulnerable populations, utilizing both algorithmic strategies and caregiver interventions [115]. Additionally, the use of advanced technologies such as natural language processing and data mining can aid in identifying and removing online content that is harmful, provocative, or lacking scientific validity [115]. It is also important to keep in mind the positive aspects of social media use, including the increased opportunities it allows for social connection [5]. Altogether, efforts to ensure safe social media for adolescents are crucial, as highlighted also by the development strategies undertaken in Europe [24, 25].