Introduction

Research background

Safety perceptions are cognitive response triggered when individuals perceive external danger and feel threatened (Bar‐Tal and Jacobson 1998). Those with lower perceived safety exhibit poor social adaptability (Ye et al. 2021), tend to restrict their range of activities, and are less likely to engage in physical activity, especially at night, which subsequently impacts their participation in social affairs (Foster and Giles-Corti 2008; Tandogan and Ilhan 2016). Furthermore, significant gender differences may exist in the evaluation and perception of environmental safety. Compare to males, females are more likely to perceive higher levels of criminal activity and reduced social cohesion, thereby diminishing their perceived safety (Johansson and Haandrikman 2023). These differing perspectives on safety between genders may pose challenges to policy formulation and enlarge gender conflicts, necessitating a deeper investigation into the mechanisms that shape safety perceptions. The widespread use of social media has been found to negatively impact people’s physical and mental health (Alonzo et al. 2020; Lee et al. 2022). In particular, social media depictions of crimes greatly increase generalized fear of crime (Zeng and Wong 2023) and reduce individuals’ safety perceptions. However, there remains a lack of research in this area, and further exploration is needed to understand the relationship between social media engagement and safety perceptions, including the underlying mechanisms and gender differences.

Literature review

In the past, individuals’ assessments of environment safety were significantly shaped by personal experiences and the information surrounding them. For example, those who perceive themselves as vulnerable to criminal victimization exhibit heightened fear of crime (Singer et al. 2019). Even people who seldom go out may regard the external world as unsafe if their acquaintances or neighbors express negative perceptions of society (Allik and Kearns 2017). Moreover, the Internet has become an essential source of individuals’ awareness of global developments and current social issues. As of December 2023, China has 1.092 billion Internet users, with a penetration rate of 77.5 and 97% of these users engage in online entertainment and communication (The 53rd statistical report on China’s internet development 2024). The information available to people, especially through media channels, largely influences their safety perceptions.

Social media serves as a comprehensive term for digital platforms that facilitate the creation and spread of user-generated content, including web platforms (e.g., Facebook, microblogs), online forums, and instant messaging applications (e.g., WhatApp and Wechat) (Gunitsky 2015; Zeng and Wong 2023). It provides a channel for individuals to discuss current events, which can subsequently alter their opinions and beliefs. As proposed by cultivation theory, the content presented on social media significantly influences and shapes people’s perceptions (Morgan and Shanahan 2010). Compared to traditional media like newspapers and television, social media platforms such as Facebook and WeChat offer a broader spectrum of information, including negative information (Zhang et al. 2013). Negative stimuli tend to attract more attention and more likely to be shared and spread (Fay et al. 2021). Exposure to information on social media that threatens personal safety can provoke increased concern for one’s own safety. In particular, awareness that criminal or suicidal events may lead to copycat behaviors greatly increases feelings of threat and fear of crime (Ueda et al. 2017). Furthermore, numerous studies have confirmed the correlation between social media consumption and heightened levels of fear (Intravia et al. 2017; Wu et al. 2019; Zeng and Wong 2023). Consequently, social media amplifies the visibility of dangerous criminal incidents, potentially heightening safety concerns.

Perceived safety is also influenced by social support. Perceived social support refers to the extent to which individuals receive assistance or understanding from family, friends and other significant others when they require help (Zimet et al. 1988). It not only enhances life satisfaction and reduces loneliness (Yue et al. 2023), but also mitigates the effects of depression on the quality of life of older adults with physically unwell (Kong et al. 2019). Moreover, studies have found that individuals with higher levels of social support perceive greater levels of safety, thereby reducing the incidence of dating violence (Jankowiak et al. 2020; Valente and Crescenzi-Lanna 2022). They also typically possess a stronger sense of self-efficacy (Luo et al. 2023) and exhibit greater confidence in facing external challenges, leading to increased safety perceptions. When social media promotes social support, it is associated with more positive coping strategies and lower threat assessments (Cho et al. 2023). Therefore, social support may enhance safety perceptions.

Various social media platforms have facilitated communication across temporal and spatial boundaries, enriching and sustaining relationships while enhancing social connections. Individuals can maintain contact with family, friends, and acquaintances through social media, freely disseminating information and engaging in instant interactions, thereby bolstering their sense of belonging and identity (Van Bavel et al. 2024), which in turn enhances their perceived social support (Häusser et al. 2023; Penfold and Ogden 2023). Longitudinal studies demonstrate that social media engagement enhances social support among older adults, alleviating loneliness (Zhang et al. 2021). Conversely, problematic social media use diminishes social support, negatively affecting well-being and mental health quality of life (Lin et al. 2021). In addition, research indicates that consuming information on social media can influence safety perceptions, but not necessarily perceived social support; conversely, active participation through posting and commenting enhances perceived social support (Yang et al. 2020). Consequently, social media engagement may indirectly influence safety perceptions through perceived social support.

People’s trust in others and their surroundings significantly impacts their perception of safety. General trust, the belief in the inherent goodness of human nature and not limited to a specific entity (Yamagishi and Yamagishi 1994), is crucial for the well-being of the elderly. This trust positively predicts physical well-being (Sato et al. 2018) and is associated with lower levels of depression and anxiety (Lin et al. 2021). Compared to individuals with lower general trust, those with higher general trust perceive lower risk (Siegrist et al. 2021) and tend to perceive the external environment as safer. General trust is also believed to indirectly reduce suicidal ideation by alleviating insomnia and fear of COVID-19 (Lin et al. 2022). Trust in others and the police has been found to reduce the fear of crime (Valente and Vacchiano 2021). In addition, Trust Transfer Theory posits that individuals exhibit varying degrees of trust in specific objects due to differences in their characteristics and familiarity with them, and that trust can be transferred from a known entity to an unknown one (Stewart 2003). Trust in political institutions enhancing individuals’ safety perceptions (Spadaro et al. 2020). Moreover, research suggests a reinforcing relationship between general trust and political trust (Bargsted et al. 2023), which may extend to safety perceptions. Collectively, these studies suggest that general trust significantly influences an individual’s perceived safety.

Given the diverse and often inconsistent quality of information sources on social media, skepticism about their reliability may arise, subsequently affecting general trust. For example, depictions of norm-violating behaviors, such as robbery, on social media can diminish general trust (Van Lange 2015). Problematic social media use has been found to negatively impact well-being and psychological health by undermining general trust and social support (Lin et al. 2021). However, Social Capital Theory proposes that individuals can use social networks not only to gain support and resources, but also to cultivate trust with others (Fulkerson and Thompson 2008). Study has identified a positive correlation between social media use and perceived general trust (Sun et al. 2023). During the COVID-19 pandemic, social media significantly enhanced individuals’ social trust (Liu and Yang 2021; Pan and Luo 2022). Moreover, longitudinal studies indicate that social media can mitigate the symptoms of post-traumatic stress disorder (PTSD) in individuals with higher levels of general trust (He et al. 2018). Therefore, social media engagement may indirectly influence perceived safety through its impact on general trust.

Furthermore, individuals’ attitudes towards the external environment can affect the safety perceptions. Beliefs in a just world (BJW) reflect the conviction that people ultimately receive what they deserve. This concept is divided into general belief in a just world (GBJW) and personal belief in a just world (PBJW) (Lipkusa et al. 1996). Individuals with PBJW attribute a greater sense of justice to the world than those with higher GBJW (Dalbert, 1999). Such beliefs not only enhance overall well-being but also alleviate depressive symptoms (Jiang et al. 2016). According to the Just World Belief Theory, belief in a just world enhances predictability and control over life (Lerner 1980), thereby boosting confidence in facing difficulties. Conversely, Individuals with lower levels of PBJW experience a lower sense of control (Fischer and Holz 2010) and are more likely to feel incapable of coping with external difficulties. Moreover, individuals with higher PBJW perceive lower levels of individual and group discrimination (Choma et al. 2012), have better protection against external risks (Dalbert and Donat 2015), and are more likely to view the external environment as safe (Ma et al. 2022). Thus, individuals with higher levels of PBJW also report higher levels of perceived safety.

In recent years, social media has become a powerful tool for victims to voice their grievances and for individuals to assertively defend their rights and interests, ensuring that perpetrators are held accountable (Malik et al. 2022). This mechanism significantly upholds the sense of justice for both victims and the public. Similarly, political institutions also use social media to publicize violations of social norms and beliefs, imposing appropriate penalties (Lee and Wei 2021). Interactions on social media platforms can shape individuals’ attitudes toward life events, thereby influencing their PBJW (Shin and Hampton 2023). Moreover, the dissemination of violent narratives on social media has been found to amplify negative cognitive biases and strengthen beliefs in a just world (Naderer et al. 2024). Consequently, social media engagement may also indirectly influence safety perceptions through PBJW.

In summary, social media may indirectly affect perceived safety through social support, general trust, and personal belief in a just world. The vulnerability model suggests that perceived levels of fear increase when individuals feel unable to cope with external threats (Hale 1996). Due to physiological differences, women and older adults may perceive themselves as less capable of handling danger, thus feeling more vulnerable and insecure (León et al. 2022; Wu et al. 2019). Further research indicates that males tend to feel safer due to social media engagement, while females experience the opposite effect (Kong et al. 2023). This may be linked to their motivations for using social media; men often use it for gaming and making new social connections, whereas women use it more for maintaining established close relationships (Manago et al. 2023; Muscanell and Guadagno 2012), making men’s social support more susceptible to social media influence. Furthermore, the impact of social media on beliefs can affect safety perceptions. Trust in men’s social media competence was found to influence general trust in members more significantly than women’s (Sun et al. 2018), suggesting gender differences in the impact of social media on general trust. Compared to women, men with higher beliefs in a just world tend to exhibit lower tendencies toward aggressive behavior (Sang et al. 2023). Thus, gender moderate the effects social support, general trust, and personal belief in a just world.

Research gaps

Although cultivation theory suggests that social media engagement shapes and influences individuals’ perceptions and beliefs (Morgan and Shanahan 2010), this process has not been explored in depth. Previous researches have focus on the aspects related to fear of public space (Almanza Avendaño et al. 2022), fear of crime (Custers et al. 2017), cyberbullying (Rasheed et al. 2020). However, the impact of the online world on real-life safety has not been thoroughly explored. Addressing this issue is crucial for understanding the mechanisms through which social media affects real-life perceptions of safety and for developing strategies to enhance these perceptions. Additionally, women generally report feeling less safe compared to men (León et al. 2022). Given the distinct motivations that men and women have for using social media, there may also be gender differences in how social media influences perceptions of safety. Clarifying these differences can help in tailoring strategies to enhance safety perceptions for both genders. Therefore, an in-depth exploration of the mechanisms and gender differences in the impact of social media on safety perceptions is necessary.

Network analysis is extensively employed in psychiatric research and has recently been increasingly integrated with structural equation modeling to clarify complex relationships within models (Liu et al. 2018). Network analysis consists of edges and nodes, where edges represent relationships between variables and nodes are evaluated using centrality indices, including strength, closeness, and betweenness (Hevey 2018). Strength indicates the tendency of a node to occupy a central position within the network, representing the cumulative sum of weighted interconnections between nodes. Closeness is the inverse of the cumulative sum of the shortest paths connecting nodes, while betweenness measures the frequency with which a node appears on the shortest route between other nodes (Cai et al. 2020). The stability of a network node is expressed by the correlation stability (CS) coefficient, with stability considered robust when the CS coefficient exceeds 0.5 (Epskamp et al. 2018).

The current study

Individuals’ assessment of environmental safety significantly shapes behavioral responses and interpersonal interactions. Traditionally, evaluations of societal security were based on personal experiences and direct interactions. However, in the digital era, social media has revolutionized access to information, particularly regarding incidents of violence and crime. The portrayal of such events in the media influences how individuals perceive overall societal safety. To elucidate these issues, this study employs network analysis and structural equation modeling to investigate the roles of social media engagement and safety perceptions from both macro (personal belief in a just world) and micro perspectives (general trust and perceived social support), analyzing the key factors that shape safety perceptions. Furthermore, this study examines the influence of gender on these processes through multi-group comparisons. Consequently, we propose five research hypotheses (see Fig. 1):

Fig. 1: Assumed model.
figure 1

SME social media engagement, SP safety perceptions, PSS perceived social support, GT general trust, PBJW personal belief in a just world.

Full size image

Hypothesis 1: Social media engagement negatively predicted with an individual’s safety perceptions.

Hypothesis 2: Perceived social support serves as a mediating factor in the relationship between social media engagement and safety perceptions.

Hypothesis 3: General trust plays a mediating role in the relationship between social media engagement and safety perceptions.

Hypothesis 4: Personal belief in a just world serve as a mediating role in social media engagement and safety perceptions.

Hypothesis 5: Gender moderates the impact of social media engagement in indirectly influencing safety perceptions through perceived social support, general trust, and personal belief in a just world.

Method

Participants

The study recruited 1481 participants using a Chinese online platform (https:// www.wjx.cn/), and the entire study was approved by the Ethical Committee of Fujian Normal University. For data preprocessing, the data of participants who did not consent to participation in the study (94 respondents), were aged under 18 years (18 respondents), or exhibited undue promptness were excluded (21 respondents). Ultimately, a total of 1348 valid participates (91%) were obtained, including 938 females (Mean age = 19.40, SD = 1.23).

Measures

Perceived social support (PSS)

Perceived social support scale was developed by Zimet et al. (1988) to assess an individual’s perceived levels of social support. The reliability of the scale has been validated in previous studies (Xiang et al. 2020). It comprises three dimensions: family, friends, and others, each dimension contains four items. Examples include “I can talk about my problems with my family” for the family dimension, “I have friends with whom I can share my joys and sorrows” for friends, “There is a special person in my life who cares about my feelings” for social support from others. Participants are required to rate 1 (totally disagree) to 7 (totally agree) on 12 items. Higher scores indicate greater perceived social support. The Cronbach’s alpha in this study was 0.97.

Social media engagement (SME)

The Chinese version of the social media engagement scale (SMES-A) was used as a tool to measure individuals’ perspectives regarding their use of social media (Ni et al. 2020). This scale comprises ten items, encompassing cognitive, affective, and behavioral aspects. The affective and cognitive domains each encompass three items, illustrated by statements like “Compared to the real world, social media makes me feel more comfortable” for the affective domain and “The support and encouragement of others on social media is very important to me.” for the cognitive domain. The behavioral facet comprises four items, exemplified by the statement “I browse social media whenever I have time.” Participants are required to evaluate each item on a 5-point Likert scale, where 1 = “strongly disagree” and 5 = “strongly agree”. Higher scores indicate higher levels of social media engagement. The Cronbach’s alpha was 0.90 in this study.

General trust (GT)

The General Trust Scale serves as a valuable tool for evaluating individuals’ overall trust levels, as proposed by Yamagishi and Yamagishi (1994). The Chinese version of this scale has been validated in prior research (Shao et al. 2019). This scale encompasses six items, including statements such as “Most people are basically good-natured and kind”. Each item requires participants to score on a 5-point Likert scale (1 = “strongly disagree” and 5 = “strongly agree”). Higher scores indicate higher levels of general trust. In this study, Cronbach’s alpha was 0.95.

Personal belief in a just world (PBJW)

Personal belief in a just world was measured using Belief in a Just World scale developed by Dalbert (1999), and the Chinese version of the scale has been validated by previous studies (Jiang et al. 2016). Participants are required to rate 1 (totally disagree) to 7 (totally agree) on seven items (e.g., “I believe that most of the things that happen in my life are fair.”). Higher scores indicate stronger personal belief in a just world. In this study, Cronbach’s alpha was 0.96.

Safety perceptions (SP)

Safety perceptions measure an individual’s feelings of safety while walking alone during the daytime and at night (Rountree and Land 1996). It comprises two items: “How safe do you feel walking alone in night time?” and “How safe do you feel walking alone in day time?”. Participants were required to perform a Likert 5-point scale (1= not saft at all, 5 = very safe). Higher scores indicate higher levels of safety perceptions, and the Chinese version of the scale has been validated in previous studies (Zhong 2010).

Statistical analysis

The study employed R version 4.2.2 for data analysis, utilizing the bruceR package for descriptive statistics and correlation analysis. Subsequently, the lavaan package was applied for assessing common method bias, structural equation modeling, and multi-group comparisons. To examine the network structure of participants and identify key nodes, JASP 0.18.3.0 was used, while the NetworkComparisonTest package in R facilitated network comparisons between genders (Van Borkulo et al. 2023). The EBICglasso algorithm is crucial for eliminating spurious correlations in the network (Chen and Chen 2008). Items with factor loadings below 0.6 were removed. The model fit was considered to be in the acceptable range when the comparative fit indices (CFI) and the Tucker-Lewis index (TLI) were >0.9 and the root mean square error of approximation (RMSEA) was <0.08 (Martynova et al. 2018).

Common method bias

As the study required participants to complete multiple questionnaires, the potential for homoscedastic bias emerged. Consequently, the research employed a common method bias analysis, specifically the Harman single-factor CFA test (Podsakoff et al., 2012). This analysis involved fitting all variables onto a single factor. The results of this analysis indicated that the fitness of single-factor model was poor (χ² = 18318.23, df = 594, χ²/df = 30.84, CFI = 0.63, TLI = 0.61, RMSEA = 0.15), indicating that common method bias had little effect on the results of the study.

Result

Descriptive statistics and correlation analysis

Table 1 presents the results of descriptive statistics and Pearson’s correlation analysis for all variables. The results demonstrated a significant positive correlation between SME and all variables, and SP was significantly positively correlated with PSS, GT, and PBJW, respectively.

Table 1 Means, standard deviations, and pearson correlation coefficients for all variables (n = 1348).
Full size table

Structural equation modeling analysis

The study employed confirmatory factor analysis to test the measurement model. Explicit variables corresponded to the items associated with each factor, while latent variables represented the scales used in the study. The results showed good fitness (χ²/df = 4.16, χ² = 2356.10, df = 566, p < 0.001, RMSEA = 0.05, CFI = 0.96, TLI = 0.96).

The mediating effects of perceived social support, general trust and personal belief in a just world

After controlling for age, education and subjective socioeconomic status, the study employed maximum likelihood estimation to assess model fit, yielding satisfactory results (χ²/df = 6.57, χ² = 4370.82, df = 665, p < 0.001, RMSEA = 0.06, CFI = 0.92, TLI = 0.91). The result revealed a non-significant relationship between SME and SP (β = −0.02, p = 0.718). Moreover, SME positively predicted PSS (β = 0.47, p < 0.001), GTS (β = 0.52, p < 0.001), and PBJW (β = 0.47, p < 0.001). PSS was positively predicted SP (β = 0.17, p < 0.001), GT positively predicted SP (β = 0.08, p = 0.039), and PBJW positively predicted SP (β = 0.22, p < 0.001). The mediating effect values were 0.08 (95% CI = [0.067,0.170], p < 0.001) for PSS, 0.04 (95% CI = [0.003,0.113], p = 0.040) for GT, and 0.10 (95% CI = [0.101,0.208], p < 0.001) for PBJW. The results suggesting that PSS, GT, and PBJW fully mediate the relationship between SME and SP (see Table 2). Consequently, Hypotheses 2–4 were substantiated.

Table 2 The mediating effects of PSS, GT and PBJW.
Full size table

Multigroup comparison

Additionally, to test whether there are differences in the processes by which SME indirectly influences SP via PSS, GT, and PBJW across genders, the study conducted cross-gender measurement invariant analyses, including construct invariance, metric invariance, and scalar invariance. Construct invariance, which tests for equivalent measurement structures across genders, showed a satisfactory model fit (χ²/df = 2.98, χ² = 3376.55, df = 1132, p < 0.001, RMSEA = 0.05, CFI = 0.95, TLI = 0.95). Metric invariance, which constrains factor loadings to be the equal between groups, and showed favorable model fit (χ²/df = 2.92, χ² = 3391.94, df = 1163, p < 0.001, RMSEA = 0.05, CFI = 0.95, TLI = 0.95). Scalar invariance, which constrains both factor loadings and intercepts to be the equal across groups, was similarly within acceptable limits (χ²/df = 2.84, χ² = 3391.94, df = 1194, p < 0.001, RMSEA = 0.05, CFI = 0.95, TLI = 0.95). Comparisons across models showed no significant differences according to likelihood ratio tests (p > 0.05), and ΔRMSEA, ΔCFI, and ΔTLI values were all less than 0.01, suggesting that measurement invariance was achieved across genders.

In addition, after establishing measurement invariance, the study constructed both constrained and unconstrained models. The constrained model imposes restrictions on all structural coefficient, while the unconstrained model restricts only the factor loadings, allowing all other parameters can be estimated freely. The results showed that the constrained model fit well (χ²/df = 4.05, χ² = 5595.22, df = 1380, p < 0.001, RMSEA = 0.07, CFI = 0.91, TLI = 0.91), and the free estimation model fit well (χ²/df = 4.08, χ² = 5557. 24, df = 1361, p < 0.001, RMSEA = 0.07, CFI = 0.91, TLI = 0.91), but the χ² likelihood ratio test was significant for both models (Δχ² = 37.98, Δdf = 19, p = 0.006), indicating that gender had a significant moderating effect on model structure. Stepwise constraints and releases for each pathway ultimately revealed that no gender differences were found only in the two pathways between social media engagement and personal belief in a just world regarding safety perceptions (p > 0.05). It is noteworthy that gender influenced the research paths differently. Specifically, the effects of total SME on SP varied between males and females. Males showed an effect primarily through the indirect paths of PSS and PBJW, whereas females also demonstrated an effect through the indirect path of GT. In addition, the predictive effects of PSS, GT, and PBJW, as well as the predictive effect of PSS on SP, were stronger for males compared to females (see Fig. 2).

Fig. 2: Multiple comparisons between males and females.
figure 2

In the figure, each line represents two groups, with the former indicating males and the latter representing females. *p < 0.05, **p < 0.01, ***p < 0.001; SME social media engagement, SP safety perceptions, PSS perceived social support, GT general trust, PBJW personal belief in a just world.

Full size image

Network analysis

Network analysis was performed for all variables by dimension across genders, as shown in Fig. 3. There are 9 nodes for both males and females, 26 non-zero edges for males and 27 for females. For males, the network structure revealed the strongest edge connections for SME2 and SME3 (0.56), while for females, it was PSS2 and PSS3 (0.60). In addition, SME2 and SME3 negatively correlated with PSS1 among males (−0.05, −0.03), and SME3 was also negatively correlated with PBJW and SP (−0.06, −0.10). Among females, SME3 was negatively correlated with PSS3, PBJW, and SP among females (−0.08, −0.02, −0.05), and SME1 was also negatively correlated with GT (−0.02). Regarding SP, the strongest connection was between PSS1 and SP for males (0.22), with no comparable connection observed for GT. In contrast, for females, the connection between PBJW and SP exhibited a greater weight (0.13). As shown in Fig. 4, the centrality test reveals that for females, PSS3 is the most central node with the highest strength (1.44). The correlation stability (CS) coefficients for the centrality indicators showed better stability for strength, with coefficients of 0.67 for males and 0.75 for females. In addition, a Network Comparison Test (NCT) was conducted to evaluate differences in network structure between genders. The results revealed significant differences in network structure (M = 0.22, p = 0.002) and overall strength (s = 0.43, p = 0.026). Specifically, the overall strength of the male network (global strength = 4.37) was significantly higher than that of the female network (global strength = 3.94).

Fig. 3: EBICglasso network structure across genders.
figure 3

a EBICglasso network structure for males and b EBICglasso network structure for females. PSS1 perceived social support from the family, PSS2 perceived social support from the friends, PSS3 perceived social support from the others, SP safety perceptions, SME1 social media behavioral engagement, SME2 social media cognitive engagement, SME3 social media emotional engagement, PBJW personal belief in a just world, GT general trust.

Full size image
Fig. 4: Indicators of centrality of network nodes for males and females.
figure 4

PSS1 perceived social support from the family, PSS2 perceived social support from the friends, PSS3 perceived social support from the others, SP safety perceptions, SME1 social media behavioral engagement, SME2 social media cognitive engagement, SME3 social media emotional engagement, PBJW personal belief in a just world, GT general trust.

Full size image

Discussion

This study examines the relationship between social media engagement and safety perceptions, focusing on its underlying mechanisms. The findings reveal that social media engagement significantly increased safety perceptions, but this process is fully mediated by perceived social support, general trust, and personal belief in a just world. Notably, the study also found that the impact of social media engagement on the mediating variables is more pronounced among males, although general trust does not significantly influence their sense of safety.

The relationship of social media engagement and perceived social support

This contrasts with prior research indicating that consumption of social news through traditional media, such as television and the internet, is associated with higher levels of fear of crime (Wu et al. 2019), and that exposure to social media such as WeChat heightens generalized fear and decreases perceptions of safety (Zeng and Wong 2023). This effect is particularly pronounced when individuals encounter portrayals of suicides on social media, which can exacerbate anxiety and suicidal tendencies (Ueda et al. 2017). Nonetheless, the findings of this study demonstrate that social media engagement can enhance perceived safety, with this effect being mediated exclusively through perceived social support, general trust, and personal belief in a just world. Social media facilitates the establishment of social relationships, which are fundamental to human needs for belonging and security (Houghton et al. 2020). Notably, while general social media use did not influence risk perception, selective exposure to social media increased perceived risk, with higher media source credibility correlating with reduced perceived risk (Niu et al. 2022).

The way people evaluate safety

The findings indicate that social media engagement enhances social support and indirectly increases perceptions of safety, which is consistent with previous research. Interactive communication on social media not only bolster individuals’ ability to maintain relationships but also elevate their sense of belonging and identity (Van Bavel et al. 2024), thereby enhancing social support (Häusser et al. 2023; Penfold and Ogden 2023). While problematic social media use weakens social support (Lin et al. 2021). In addition, social support is associated with improved physical and mental health in individuals under 60 (Park et al. 2023), potentially providing additional resources for addressing external threats and challenges. Beliefs about social support not only promote self-efficacy and increase an individual’s ability to cope with danger (Luo et al. 2023), but is also associated with more positive coping styles and lower threat assessments (Cho et al. 2023). Consequently, social media engagement can indirectly enhance safety perceptions through social support.

Research indicates that social media engagement enhances general trust, aligning with prior findings (Sun et al. 2023). Social networks not only provide individuals with support and resources but also foster trust (Fulkerson and Thompson 2008; Lin et al. 2021). This may be attributed to social media’s capacity to offer users greater control over accessing detailed information about relevant events, thereby enhancing general trust (Seo et al. 2012). Higher levels of general trust are associated with better physical and mental health (Lin et al. 2021; Sato et al. 2018), and lower perceived risk (Siegrist et al. 2021). Trust Transfer Theory suggests that trust in a specific object can be transferred from a known target to an unknown one (Stewart 2003). Similarly, research has demonstrated a reciprocal relationship between general trust and political trust (Bargsted et al. 2023), where trust in political institutions enhances individuals’ safety perceptions (Spadaro et al. 2020). Moreover, social media engagement and online discussions can reduce perceived risk by enhancing social trust (Liu and Yang 2021). Consequently, social media engagement can indirectly increase safety perceptions by promoting general trust.

Individuals inherently believe that wrongdoers should face consequences, and events that contradict this belief foster perceptions of an unjust world (Dalbert 1999). Research suggests that frequent social media use can bolster an individual’s beliefs in a just world, especially when strong relationships are cultivated through these platforms (Shin and Hampton 2023). Victims vocalizing their experiences on social media can enhance perceptions of justice by ensuring criminals receive due punishment (Malik et al. 2022). Similarly, political institutions use social media to publicize and punish violations of the social norms (Lee and Wei 2021), reinforcing beliefs in a just world. Such beliefs provide individuals to a sense of control and predictability in life (Lerner 1980), thereby increasing their confidence in facing difficulties. Higher personal belief in a just world enhances well-being and alleviates depressive symptoms (Jiang et al. 2016), while lower personal belief in a just world associates with low levels of perceived control (Fischer and Holz, 2010). In addition, those with higher personal belief in a just world tend to perceive lower levels of discrimination (Choma et al. 2012), lower perceived risk (Dalbert and Donat 2015), and a safer external environment compared to those with weaker beliefs (Ma et al. 2022). Thus, social media can indirectly increase an individual’s safety perceptions by elevating personal belief in a just world.

The difference on safety perceptions between genders

Furthermore, the study revealed that males perceived higher levels of social support, general trust, and personal belief in a just world in their social media engagement compared to females, though their general trust did not influence perceived safety. This disparity may be related to the differing motivations for social media use between genders. Males are inclined to build new relationships, form friendships, and engage in online gaming, while females tend to focus on maintaining existing close relationships (Manago et al. 2023; Muscanell and Guadagno 2012). Network analyses confirm that social media engagement enhances male’s familial connections, with this family support bolstering their safety perceptions. Consequently, social support plays a more substantial mediating role in the relationship between social media engagement and safety perceptions for males than for females.

Moreover, male’s trust in social media’s competence impacts general trust more profoundly than female’s (Sun et al. 2018), explaining the stronger prediction of general trust through social media for males. This aligns with network analysis findings that show a positive correlation between cognitive engagement in social media and males’ general trust. This divergence may root in evolutionary roles, with females typically managing child-rearing and being more cautious towards external information, while males are more prone to risk-taking and teamwork (Geary 2021), making trust in social media competence more valuable to them. However, general trust in others does not significantly affect men’s safety perceptions. This possibly due to their greater resilience to frustration compared to women (Wang 2020), and their reliance on self-trust for safety. Notably, men are more likely than women to report incidents of cyberbullying and victimization on social media (Donat et al. 2023), reinforcing their personal belief in a just world. Thus, while social media engagement strongly influences male’s perceived social support, general trust, and personal belief in a just world, but their safety is primarily influenced by social support and personal belief in a just world.

Limitation and future research

Nevertheless, this study has several limitations. First, the cross-sectional design was unable to reveal the causal relationship between the variables; future research could adopt a longitudinal approach to reveal more clearly the interactive process between social media engagement and safety perceptions. Second, this study focused on the overall effects of social media engagement on safety perceptions without distinguishing the impacts of different social media platforms; subsequent research could examine the distinct impacts of different social media platforms. Finally, this study did not ensure an equal representation of males and females, potentially leading to measurement errors that could influence the findings. Future studies should consider addressing these limitations to enhance the robustness of the results.

Conclusion

The study explored the influence of social media engagement on safety perceptions, highlighting the fully mediating role of perceived social support, general trust, and personal belief in a just world. Furthermore, there were gender differences in this process. Males’ beliefs being influenced by social media engagement than females’, and their safety perceptions primarily influenced by social support and personal belief in a just world. This research provides a valuable reference for understanding the mechanisms through which social media enhances individuals’ real-life safety perceptions, addressing a gap in the literature regarding the impact of social media on perceived safety. In addition, attention should shift towards how social media affects personal beliefs, rather than focusing solely on outcomes like personal health. Social media platforms should prioritize fostering and maintaining positive social support networks, while spread positive content to bolster users’ trust and belief in a just world, ultimately enhancing their safety perceptions.