Latest Issue

While surveillance can be conducted by various entities, including friends, family, employers, police and governmental bodies (Kruse et al., 2018), this study specifically focuses on government-related aspects. Government surveillance on the Internet aims to induce obedience among indiv iduals who are particularly resistant to the regime by instilling a sense of apprehension regarding the potential consequences of engaging in dissenting political activities online (Krueger, 2005). Some governments can track and analyze online activities and prosecute social media users who spread opposing views (Feldstein & Wong, 2020). Online surveillance involves gathering, storage, and examination of digital footprints by governmental bodies, with potential targets ranging from individuals to entire societies (Chan et al., 2024).

Researchers of political communication have regarded online surveillance as a significant area of study. Previous research has explored individuals’ reactions and responses to surveillance, but has found inconsistent patterns (e.g., Krueger, 2005; Pan & Siegel, 2020; Stoycheff, 2016). Individuals alter their behaviors when they become aware of the presence of surveillance (Kruse et al., 2018). Docility and resistance are two general reactions to surveillance (Krueger, 2005). Regarding docility, by instilling fear and increasing the risk of participating, surveillance can “reduce individuals’ propensity to dissent” (Hager & Krakowski, 2022, p. 565). Some studies have suggested that individuals’ inclination to express themselves on online platforms may be curtailed because of government surveillance (e.g., Duffy & Chan, 2019; Oz & Yanik, 2022; Staples, 2013). One study targeting internet users in the US revealed that eighty percent of the respondents would be “much less likely” or “somewhat less likely” to speak or write about certain topics online and be more careful about what they say under government surveillance online, indicating the potential chilling effect on speech (Penney, 2017). When surveillance is recognized and internet users can communicate with each other while observing the watchers, resistance can arise (Krueger, 2005). The concept of a backlash effect (Sullivan & Davenport, 2017) finds support in decades of research on repression. For instance, Lemaire (2023) pointed out that a backlash effect is more likely to be evoked if the surveillance is too obvious and directly experienced by citizens (e.g., shutting down the whole social media platform).

The form and structure of a political regime play a crucial role in understanding the impact of government surveillance on online political participation. Individual responses to surveillance may vary significantly depending on several factors, including the level of political and media freedom, the prevailing political culture, and the robustness of civil society within a given context. A cross-national study by Kalmus et al. (2022) provides valuable insights into this complex relationship. The researchers examined individuals’ responses to surveillance across three countries with distinct political histories: Estonia, Portugal, and Sweden, respectively. These nations were selected to represent past totalitarian, authoritarian, and long-standing liberal democratic regimes, respectively. The study’s findings revealed an intriguing pattern: in Portugal, where trust in state institutions and media is the lowest among the three countries, citizens exhibit the least tolerance for online state surveillance.

Despite these specific findings, it is crucial to note that citizens’ reactions to surveillance demonstrate significant variability across countries with different regimes, with no consistent patterns emerging. The chilling effect can occur across different political systems. In repressive regimes like Russia, the government may cultivate an environment where citizens come to accept censorship and even view surveillance as beneficial (Nisbet et al., 2017). Meanwhile, in democratic states like the U.S., the 2013 revelation of government surveillance practices also led to a chilling effect on online search behavior (Marthews & Tucker, 2015).

Likewise, a backlash effect can arise under different regimes, even in authoritarian contexts (e.g., Moore-Gilbert & Abdul-Nabi, 2021; Pan & Siegel, 2020). In China, public outrage over the Green Dam Youth Escort filtering policy sparked protests that temporarily stopped its implementation, illustrating that even in repressive political environments, citizens will push back when their privacy and freedom are visibly harmed (Monggilo, 2016). Similarly, in Burundi, the disruption of WhatsApp led to a surge in the use of alternative tools to resist surveillance (Falisse & Nkengurutse, 2019).

These examples suggest that it is not the regime type that primarily shapes citizens’ reactions to surveillance, but rather the visibility and direct impact of such measures. When surveillance is highly visible, and citizens feel its direct effects, backlash often follows. In the context of the Internet, government surveillance is more exposed, and online platforms offer citizens greater flexibility to resist repressive measures. This digital environment, which facilitates lowcost mobilization and participation, further contributes to backlash and more forms of online political acts against government repression (Pan & Siegel, 2020).

Any voluntary and nonprofessional activities concerning government, politics, or the state can be considered as political participation (Van Deth, 2014). The list of such participatory activities has become virtually infinite, including voting, demonstrating, contacting public officials, boycotting, posting blogs, volunteering, joining flash mobs, signing petitions and so forth (Van Deth, 2014). Scholars have classified types of political participation in various ways: individual vs. collective (e.g., Pattie et al., 2003); formal vs. extra-parliamentary (e.g, Ekman & Amnå, 2012); formal vs. informal (e.g., Jensen, 2013); institutional vs. non-institutional (e.g., Dalton, 2008); or online vs. offline (e.g., Oser et al., 2013; Vissers & Stolle, 2014). While early empirical work on political participation tended to focus mostly on voting and election-related activities, various forms of digital platforms have widened the repertoire of participatory actions and measures of e-participation (Gibson & Cantijoch, 2013). Social media platforms provide a domain of empowerment for youth, presenting novel avenues for them to express political dissent (A. Lee, 2018; Lemaire, 2023). New digital platforms provide citizens with low-cost routes to consume, produce, distribute, and comment on news and political information (Weeks et al., 2017). On the flip side, technologies have also been increasingly used by governments to monitor citizens’ browsing behaviors and online communications as well as censor information that challenges governmental policies (Datt, 2021; A. Lee, 2018).

Visibility may be a crucial factor in investigating the impact of government surveillance on individuals’ political participation in the online realm. Visibility in mediated communication is about “calling attention to something by showing it” (Dayan, 2013, p. 146). In general, the visible is something that can be seen and perceived visually, while invisible is what cannot be seen, or is imperceptible or hidden from view (Thompson, 2005). In the digital world, individuals can make their political activities visible by uploading and distributing political content or leaving comments on existing resources. They can also engage in relatively invisible ways, such as reading news or consuming political content online, which allows them to access political information without revealing their identities. Together with benefiting from the visibility affordances of social media, individuals can observe others’ political activities, which were once invisible, through social media feeds (D. H. Kim & Ellison, 2022). However, individuals have little control over how their information is distributed through the platform’s interface (Thompson, 2005; Trottier, 2011). Posts or comments made online are visible worldwide, and various entities, including friends, family, employers, law enforcement agencies, and the platform itself (Trottier, 2011), collectively contribute to a virtual ecosystem of “imagined surveillance” (Duffy & Chan, 2019, p. 121). Consequently, these acts of scrutiny give rise to a new surveillance culture, integrating surveillance in everyday life (Lyon, 2017). The greater the visibility of one’s online political activities, the higher the risk of exposure to online government surveillance.

Therefore, this study argues that the anonymity and publicity of political content that individuals publish or engage with are important factors to consider when participating in online political activities in a digital surveillance environment. Defining and categorizing visibility levels for online political participation may help develop a better understanding of the underlying mechanism between online political participation patterns under government surveillance.

According to Acock and Scott (1980, p. 67), high-visibility participation is defined as “active behaviors that can be observed or become known to others,” such as wearing a campaign button, working for a political candidate and so on. Lowvisibility participation refers to passive and nonpublic behaviors such as listening to political shows or radio. Based on previous research (e.g., Marthews & Tucker, 2017; Oz & Yanik, 2022), this study expects individuals to exercise more caution towards online political activities to avoid detection and arrest within the scope of surveillance. This caution is likely to lead to a higher degree of decline in higher visibility online political activities compared with low-visibility activities, which are perceived as relatively safer forms of participation. Therefore, we propose the following hypothesis.

• H1: Perceived online surveillance will significantly decrease high-visibility political participation more than low-visibility political activities.

Surveillance has been extensively theorized to alter human experiences by evoking emotions such as anxiety, fear, and suspicion (Ellis et al., 2013). One of the goals of surveillance is to create political isolation by using the fear of millions of people (Starr et al., 2008). Spaces generated by surveillance can be characterized as intricate and ambiguous, incorporating features that are simultaneously ever-present and unnoticed, tangible yet intangible, and even extending beyond geographical boundaries (Ellis et al., 2013). Especially in the digital era, online political contents sometimes spread beyond the intended privacy settings (boyd, 2010). Given the enduring nature of content and its ability to remain searchable over time, individuals increasingly feel uncertain about online surveillance (boyd, 2010; Thorson, 2014).

According to the cognitive appraisal theory, individuals react to stressors or events differently, leading to the experiences of various emotions such as joy, hope, sadness, anger, and fear (Vasilopoulos et al., 2019). Individuals would further take deliberate actions to cope with their emotions (Lazarus, 1991; Valentino et al., 2011). Government surveillance is linked to psychological distress, which in turn evokes various negative emotions such as fear, anxiety, and anger (O’Connor & Jahan, 2014). Hager and Krakowski (2022) noted that some individuals experience fear while others feel anger as a response to surveillance; this divergence in emotional responses may lead to varied changes in political participation. Anger, a negative emotion, is typically triggered when individuals perceive others violating commitments or transgressing rules (Best & Krueger, 2011).

Fear, another negative emotion, can prompt changes in individuals’ political behavior during uncertain and uncontrollable situations (e.g., Lerner & Keltner, 2001; Nabi, 2002; S. M. Kim & Min, 2021). Anger is known to be linked with a propensity for activism (e.g., Lerner & Keltner, 2001). Previous studies have suggested that anger positively affects political engagement, including involvement in election activities (Valentino et al., 2011), contacting politicians, and signing petitions (Best & Krueger, 2011), while fear is typically associated with avoidance behavior (e.g., Lazarus, 1991; Weber, 2013). However, other studies have demonstrated fear’s role in motivating political participation, such as encouraging information-seeking (Wagner & Morisi, 2019) or voting and signing petitions (S. M. Kim & Min, 2021; Valentino et al., 2011). Given the inconsistent findings from previous research, the role of fear in political participation remains unclear and requires further empirical investigation.

While previous research has often examined multiple negative emotions in a single study (e.g., S. M. Kim & Min, 2021; Valentino et al., 2011), the focus of this study is on the emotion of fear in response to specific government surveillance situations. Starr et al. (2008) found that, under surveillance, individuals were afraid to discuss political issues openly and honestly within their community, and often considered abandoning their political activities. The NSL has created a climate of fear in Hong Kong (Wong & Kellogg, 2021). Beginning with Jimmy Lai and the case of the 12 Hong Kongers (F. L. Lee & Chan, 2023), the arrest cases indicate the pervasive sense of fear due to government prosecution. Individuals have expressed concerns about the potential compromise of their right to a fair trial and the prospect of facing prosecution for peaceful political activism (Wong & Kellogg, 2021). Mak et al. (2024) noted that individuals experienced fear after the implementation of the NSL, triggered by the various arrest cases related to online speech. For example, within a month of the NSL’s passage, Hong Kong police arrested four student activists for the comments they posted on Facebook.

This study aims to further explore the impact of fear in the period following the enforcement of NSL in Hong Kong, which has been in place for over three years. We anticipate that individuals are cautious with the law and their online behavior owing to the numerous arrest cases. Building on prior studies and focusing on the specific context of this research, we explore how fear mediates the relationship between perceived government surveillance and online political activities with varying levels of visibility. This leads to the following research question:

• RQ1: How does the emotion of fear mediate the effects of perceived surveillance on online political participation at different levels of visibility?

Individuals can cope with the fear generated from online government surveillance by employing various tactics. In a case study of American Muslim youth, Shresthova (2016) documented how they engaged in an ongoing and complex negotiation process under surveillance, actively segmenting, filtering, and striving to control information about themselves across their networks and beyond. Although not every user possesses the capability to employ privacy management tools, such as using Virtual Private Networks (VPN) or installing spyware (Oz & Yanik, 2022), Bodo (2015) demonstrated that the utilization of privacy tools often surges in response to government surveillance. By utilizing online privacy management tools, individuals can identify their watchers (Oz & Yanik, 2022), thus feeling more comfortable (Mak et al., 2024; Marx, 2003) and safe (Oz & Yanik, 2022). Mak et al. (2024) confirmed that through active privacy management, users can alleviate their fear of the NSL and ultimately increase their level of online political participation. Oz and Yanik (2022) also successfully demonstrated the moderating role of online privacy skills in the relationship between government surveillance and willingness to speak out on social media. Nonetheless, individuals with higher levels of digital literacy may exhibit greater caution towards potential online surveillance and be more vigilant towards online monitoring activities by the government. Consequently, they may experience greater fear in surveilling contexts compared to those with lower levels of digital literacy, as they are more sensitive to online safety and have a better understanding of the potential dangers posed by online surveillance.

Previous research has suggested that online and social media privacy management skills are frequently employed in response to government surveillance. However, whether individuals can alleviate the feeling of being surveilled by implementing proactive privacy management, or, on the contrary, whether this awareness of surveillance increases the level of fear, is largely unknown. Against this backdrop, this study explores whether online privacy management skills can moderate the process by which perceived surveillance cause fear and subsequently affects political activities with varying degrees of visibility.

• RQ2: How does the level of online privacy management skills moderate the mediating effects of fear between perceived online surveillance and political participation, at different levels of visibility?

In the context of surveillance culture (Lyon, 2017), citizens are actively engaged in conceptualizing surveillance. This involves identifying the nature of the surveillance and the surveillant, and determining how to respond (Oz & Yanik, 2022). Under all-encompassing surveillance and a climate of mutual caution, any act of sharing can be seen as a form of exposure (Ball, 2009). Consequently, how individuals perceive the trustworthiness of the people around them and the platforms they use becomes more crucial in shaping their reactions to government surveillance. Governments employ diverse tactics to enhance the effectiveness of online surveillance. For instance, the Hong Kong administration introduced a mutual monitoring system enabling citizens to report potential breaches of the NSL through a dedicated hotline and WeChat (Mak et al., 2024). This system subtly encourages citizens to monitor and accuse one another, fostering a heightened fear of disclosing digital footprints to individuals one does not entirely trust (Mak et al., 2024). The accumulation of over 380,000 individual reports since the NSL’s implementation underscores the significant impact of online reporting on shaping a culture of online surveillance (Hong Kong Police Force, 2022). Within this political framework, this study examines how social trust and trust in platforms might moderate the relationship between online government surveillance and political engagement.

We conducted an online survey with a sample of Hong Kong residents (N = 310, age ≥ 18 years) who reported active social media use. Sample size adequacy was evaluated through two complementary approaches. First, we followed established guidelines for structural equation modeling that recommend 5-10 observations per estimated parameter (Bentler & Chou, 1987). Given our model’s 62 parameters, this indicated a minimum requirement of 310 participants. Second, we conducted a Monte Carlo power analysis, a method particularly suited for complex mediation models (Schoemann et al., 2017). This analysis yielded a correlation-based power estimate of .94 for our sample size of 310, exceeding Cohen’s (1992) conventional threshold of .80.

Participant recruitment was conducted through Mansfield, a Hong Kong-based survey agency, utilizing their online panel structured according to key demographic criteria. Quota sampling was employed to ensure population representativeness, with profiling questions matching gender and age distributions to the 2023 Census and Statistics Hong Kong data. Data collection occurred between April 10-15, 2024. All study participants provided informed consent, and the study design was approved by the Institutional Review Board (IRB) at Korea University under protocol number IRB-2024-0056.

The final sample demonstrated balanced gender distribution (49.4% male) and age stratification: 25.2% aged 18-29 years, 26.4% aged 30-39 years, and 24.2% each for ages 40-49 and 50+ years. Educational attainment skewed toward higher education, with 60.3% of participants holding at least a bachelor’s degree, representing an oversampling of individuals with advanced education relative to the general population. Table 1 presents detailed sample demographics. All survey materials were presented in Cantonese to ensure linguistic accessibility for participants.

H1 proposed that perceived government surveillance would exhibit differential effects on online political participation based on visibility levels, with stronger suppression expected for high-visibility compared to low-visibility activities. To test this hypothesis, we compared two structural equation models: an unconstrained model allowing independent paths from perceived surveillance to both participation modes and a constrained model equating these paths.

Both models demonstrated robust fit according to established criteria (Collier, 2020; Hu & Bentler, 1999). The unconstrained model showed strong fit indices: χ² (16, N = 310) = 17.15, p = .376, CFI = .999, RMSEA = .015, RMR = .045. Similarly, the constrained model exhibited strong fit: χ² (17, N = 310) = 18.46, p = .361, CFI = .998, RMSEA = .017, RMR = .045. A chi-square difference test revealed no significant distinction between the models, Δχ² (1) = 1.31, p = .250. Following the principle of parsimony, we selected the constrained model, which revealed equivalent suppression effects of perceived surveillance on both high- and low-visibility political activities (b = -0.08, 95% CI [-0.15, -0.02], SE = 0.04, p = .038). These results failed to support H1’s prediction of differential effects based on participation visibility levels.

Examining the mediating role of the emotion of fear (RQ1), results showed that perceived surveillance significantly increased fear (b = 0.48, 95% CI [0.39, 0.55], SE = 0.05, p < .001), which subsequently demonstrated positive associations with both high-visibility (b = 0.12, 95% CI [0.04, 0.19], SE = 0.05, p = .025) and low-visibility activities (b = 0.13, 95% CI [0.04, 0.21], SE = 0.05, p = .012). That is, individuals experiencing greater fear under perceived government surveillance showed increased engagement in both expressive and information-seeking political activities. The indirect effects through fear were significant for both high-visibility (b = 0.06, 95% CI [0.02, 0.09], SE = 0.02, p = .014) and lowvisibility engagement (b = 0.06, 95% CI [0.02, 0.10], SE = 0.03, p = .016).

Collectively, these findings revealed a complex relationship between surveillance, fear, and political participation. While initial zero-order correlations showed no significant associations between perceived surveillance and both highvisibility (r = .00, p = .969) and low-visibility (r = .08, p = .151) political participation (Table 2), two distinct pathways emerged through path analyses. First, when controlling for fear, surveillance awareness demonstrated significant negative direct effects on both participation types, indicating fear’s role as a suppressor variable. The suppression effect manifests in how the inclusion of fear strengthens the predictive value of perceived surveillance (the initial predictor) on political participation (the outcome), compared to their zero-order correlation (Conger, 1974; Holbert et al., 2012).

Second, perceived online surveillance demonstrated a strong positive relationship with fear, which in turn positively influenced both participation types. This positive indirect effect mediated by fear counterbalanced the negative direct effect of surveillance awareness on participation, thereby explaining the negligible zero-order correlations observed in our initial analysis. As demonstrated in Table 3, these opposing mechanisms—negative direct effects and positive indirect effects—effectively neutralized each other, resulting in nearly null total effects of perceived surveillance on online political engagement across visibility levels.

For RQ2, we investigated whether online privacy management skills moderated the mediating effects of fear in the relationship between perceived surveillance and political participation across visibility levels. We incorporated an interaction term between perceived surveillance and privacy management skills into the model. Analysis revealed that this interaction term did not significantly affect fear (b = 0.03, 95% CI [-0.03, 0.09], SE = 0.04, p = .428), suggesting that individual privacy management capabilities do not moderate the relationship between surveillance and fear, thereby leaving the fearmediated pathway unaffected.

Testing the moderating effect of social trust (H2), we incorporated an interaction term between perceived surveillance and social trust into our model. The analysis revealed that social trust neither exhibited significant direct effects on highvisibility (b = 0.00, 95% CI [-0.04, 0.04], SE = 0.02, p = .969) nor low-visibility (b = -0.02, 95% CI [-0.07, 0.01], SE = 0.02, p = .387) political participation, nor moderated the relationship between perceived surveillance and political participation (high-visibility: b = -0.02, 95% CI [-0.06, 0.01], SE = 0.02, p = .288; low-visibility: b = -0.02, 95% CI [-0.05, 0.02], SE = 0.02, p = .461). Thus, H2 was not supported, indicating that the effect of perceived surveillance on online political participation remains constant across different levels of social trust.

The analysis of platform trust’s moderating role (H3) revealed a more nuanced pattern. Platform trust significantly moderated the negative impact of surveillance awareness, but exclusively for lowvisibility activities; the interaction term between perceived surveillance and platform trust showed a significant impact only on covert participation (b = 0.10, 95% CI [0.02, 0.19], SE = 0.05, p = .036). Subsequent spotlight analysis (detailed indices in Table 4) demonstrated that when platform trust was below average, perceived surveillance significantly suppressed low-visibility political activities (b = -0.13, 95% CI [-0.25, -0.02], SE = 0.06, p = .021). However, this suppression effect became nonsignificant at mean and above-average levels of platform trust, suggesting that higher platform trust can buffer against surveillance’s chilling effects on low-visibility political activities such as information seeking and news consumption online (Figure 2).

Figure 1.  Predicting Online Political Participation at Different Visibility Levels

Note. Values are unstandardized coefficient. Significant paths are indicated by solid lines; nonsignificant paths are indicated by dashed lines. Online privacy management skills, social trust, platform trust, and two control variables (political interest and age) were included in the model but omitted from the figure.

* p < .05. ** p < .01. *** p < .001.

Figure 2.  Platform Trust as the Moderator in the Relationship between Perceived Surveillance and Low-Visibility Participation

In contrast, while platform trust exhibited a significant boost effect on more visible political participation (b = 0.13, 95% CI [0.05, 0.23], SE = 0.06, p = .019), it did not meaningfully condition the effects of surveillance awareness on highvisibility political activities (b = 0.02, 95% CI [-0.07, 0.10], SE = 0.05, p = .715).

Collectively, these results provide only partial support for H3, indicating that the chilling effects of surveillance awareness on more expressive political activities persist regardless of individuals’ trust in social media platforms.