When politics is personal: Curating safe spaces through disconnection on instant messaging platforms

ABSTRACT

While existing research suggests that the closed digital spaces of Instant Messaging (IM) platforms have important democratic implications, this study examines how users actively curate such spaces through disconnection. Using a mixed-method approach within the context of Hong Kong politics after the introduction of the National Security Law (NSL), we examine the extent to which and why IM users engage in disconnective behaviors (tie dissolution, content filtration), and explain the behaviors through socio-psychological and structural factors. Statistical analysis of survey data (N = 946) shows that encountering political disagreement on IM platforms predicts disconnection, which is strengthened by fear of social isolation and perceived systemic victimhood. Thematic analysis of open-ended survey questions contextualizes these findings: Through tie dissolution, IM users regulate their own visibility and remove problematic contacts to avoid being penalized by authorities and peers. By muting disagreeable content, they turn away from contentious political discourse perceived as risky and ineffective under the NSL to focus on personal and social well-being. Overall, we argue that disconnection is a necessary means to curate safe spaces of withdrawal to cope with an uncertain and insecure political environment – it follows the logic of self-care and is a product of social and systemic threats.

KEYWORDS:

• social media
• instant messaging (IM) platforms
• disconnective behaviors
• safe spaces
• systemic victimhood
• fear of social isolation

Introduction

In recent years, we have seen an on-going trend of social media users migrating conversations from relatively open, public social network sites (SNS) to more closed and private instant messaging (IM) platforms like WhatsApp and Telegram (Newman, Fletcher, Kalogeropoulos, & Nielsen, 2019). Studies show that these IM platforms provide “safe spaces” for political talk thanks to the intimate and controlled social settings (Valeriani & Vaccari, 2018; Velasquez, Quenette, & Rojas, 2021; Zhu, Esteve-Del-Valle, & Meyer, 2022). They also host larger groups that are only accessible to group members, offering so-called “meso news-spaces”—spaces between the public and private, where members engage with news and politics through informal social interactions (Tenenboim & Kligler-Vilenchik, 2020). Many thus believe that these closed digital spaces harbor the potential to fulfill a discursive democracy. However, we know little about how IM users actively curate these spaces. This study thus fills the gap by looking at disconnective behaviors.

Disconnection is an essential element in people’s social media practices. It provides an important lens to understand how citizens (dis)engage in politics in everyday online interactions (John & Dvir-Gvirsman, 2015; Zhu & Skoric, 2022). Existing research primarily focuses on what is termed “politically motivated unfriending” on SNSs like Facebook, and mainly explains the phenomenon from the perspective of political disagreement. Following the normative expectation that contact with opposing views is critical to the deliberative processes, scholars often see disconnection as democratically problematic because it reduces political diversity (e.g., John & Dvir-Gvirsman, 2015; Neubaum, Cargnino, & Maleszka, 2021; Yang, Barnidge, & Rojas, 2017; Zhu, Skoric, & Shen, 2017). In this study, we highlight that whether, how, and why people resort to disconnection when facing political disagreement depends on the political environment they are in. To that end, we conduct the study within the context of Hong Kong politics one year after the introduction of the National Security Law (NSL).

Hong Kong has a hybrid political system. In the past decades, the territory enjoyed a culture of political activism as seen in the large waves of pro-democracy, anti-government social movements, such as the 2014 Umbrella Movement and the 2019 Anti-Extradition Bill Movement. On 30 June 2020, the Chinese central government passed the National Security Law by decree, bypassing the city’s legislature, despite large opposition from Hong Kong residents. The legislation claims to “prevent, curb and punish acts of secession, subversion of state power, terrorist activities, and collusion with foreign or external forces to endanger national security” in order to “maintain Hong Kong’s long-term stability and prosperity” (Hou & Peng, 2023, p. 2). Critics argue that the NSL eroded Hong Kong’s autonomy and political freedom, and decapitated pro-democracy movement through censorship, surveillance, and retribution: It resulted in the arrest and prosecution of prominent pro-democracy politicians and activists, closure of major independent news outlets, and disbandment of civil society organizations (Freedom House, 2022). The Hong Kong police allegedly set up hotlines to encourage citizens to monitor and inform on each other for suspected breaches of the NSL (Davidson, 2020). In sum, this study is contextualized in a political climate of uncertainty and insecurity due to the growing state and social sanctions. While state oppression of political dissent is more common in non-democratic regimes, fear of social sanctions such as social exclusion, economic retribution, and physical retaliation is not uncommon to marginalized groups in democracies that such fear plays a large role in their political participation and expression (A. Lee, 2018, 2022; Van Duyn, 2021).

Against this background, we conceptualize disconnection as a means to build safe spaces—spaces that are safe from social sanctions and systemic oppression. Combining statistical analysis of survey data and thematic analysis of open-ended questions, this study examines 1) the extent to which IM users engage in politically motivated disconnective behaviors on IM platforms, 2) their reasons and associated meanings, and 3) how fear of social isolation and perceived systemic victimhood influence disconnective behaviors in the face of political disagreement. This mixed-method approach allows us to identify patterns that are generalizable and understand them in a given political context. This study’s contribution is twofold. First, it expands the literature on politically motivated unfriending by closely examining two types of disconnective behaviors, namely dissociation and content filtration on IM platforms – a social media environment that is distinct from SNSs. Dissociation refers to severing ties with someone or a group (e.g., deleting and blocking someone, removing someone from a group, leaving a group), whereas content filtration involves hiding content while keeping the ties (e.g., muting someone, muting a group). Second, moving beyond the normative theory of the public sphere, this study advances the theoretical framework of “safe space”: We argue that people disconnect in response to what they perceive as unsafe in the political and social environment – it is a means to curate safe spaces of withdrawal to cope with social risks and systemic threats, which is necessary for participation, safety, and well-being.

Literature review

Disconnection on IM platforms: curating a closed digital environment

There is a burgeoning research interest in disconnective political behaviors on SNSs, namely unfriending and unfollowing for reasons related to politics. Commonly termed “politically motivated unfriending”, it refers to “a conscious act by a person to end the dyadic relationship and manifests itself through the removal of a link between the dyad” (Sibona, 2014, p.1677). In other words, it allows users to socially disconnect from others, and by doing so constrict the boundaries of their online networks and engage in post-hoc content filtration (Bode, 2016; Yang, Barnidge, & Rojas, 2017). Existing findings suggest that heavier SNS users with larger and more diverse networks are more inclined to sever or suspend ties with others because of political disagreement. Weak ties are more susceptible than strong ties because they bring in differing views but lack the kind of relationship qualities that prevents conflicts (Barnidge, Peacock, Kim, Kim, & Xenos, 2022; John & Dvir-Gvirsman, 2015; Neubaum, Cargnino, & Maleszka, 2021; Neubaum, Cargnino, Winter, & Dvir-Gvirsman, 2021; Zhu & Skoric, 2021). Notably, these findings apply to SNSs where political interactions are often difficult and uncertain because of the large number of weak ties, collapsed contexts, and undefined boundaries between the private and public (Thorson, 2014). However, we do not know whether, to what extent, and how people engage in disconnective behaviors on IM platforms that are distinct from SNSs.

Existing literature shows that, compared to many SNSs, IM platforms afford users high control over the boundaries of social contexts and flows of content (Karapanos, Teixeira, & Gouveia, 2016; Matassi, Boczkowski, & Mitchelstein, 2019). Sociability on these platforms is mainly produced in small groups and one-to-one chats (Boczkowski, Matassi, & Mitchelstein, 2018; Karapanos, Teixeira, & Gouveia, 2016; O’Hara, Massimi, Harper, Rubens, & Morris, 2014). They are bound to encounters and relationships in the offline world among those who are socially, emotionally, and/or geographically close, such as friends, families, housemates, and coworkers (O’Hara, Massimi, Harper, Rubens, & Morris, 2014). Among them, the spontaneous social exchange of small, inconsequential information often invites news sharing and political talk, although people navigate such conversations carefully as they perceive politics as personal and divisive (Swart, Peters, & Broersma, 2019; Zhu, Esteve-Del-Valle, & Meyer, 2022). IM platforms also host larger, interest-oriented groups, offering so-called “meso news-spaces”—spaces that are open to a wider set of diverse audience but only accessible to group members (Tenenboim & Kligler-Vilenchik, 2020). Research finds that in meso spaces dedicated to politics, sociability among members who are initially strangers makes political talk more intimate, civil, and yet socially risky at the same time (Kligler-Vilenchik, 2021).

In sum, IM platforms afford controlled and closed social settings where the dominant model of communication is personal, and the personal is intertwined with the political (Valenzuela, Bachmann, & Bargsted, 2021). While this encourages political talk, it also breeds conflicts. Personal connections can make disagreements more personal and turn political arguments into attacks perceived to be of an intimate nature. After all, “when we are most exposed and open, we are also most vulnerable” (Kligler-Vilenchik, 2021, p. 12). Indeed, research shows that conflicts arise in meso news-spaces and even among friends and family members, which sometimes leads to disconnection such as leaving a group (Kligler-Vilenchik, 2021; F. L. Lee, 2022). This accelerates in contentious times. While in everyday life people tend to value social resources, downplay the importance of politics, and overestimate political homogeneity in their social circles, in times of political conflicts, they are more eager to take sides and more often exposed to questionable views from their social contacts, which leads to feelings of shock and disappointment, relationship strains and breakdown (Bozdag, 2020; John & Gal, 2018; Schwarz & Shani, 2016). Furthermore, existing research suggests that users perceive SNSs as their own domestic places and engage in politically motivated unfriending to remove content that is not to their liking (Bozdag, 2020; John & Gal, 2018; Jordá, Cañedo, Bene, & Goyanes, 2021; Schwarz & Shani, 2016). Given that IM platforms are often experienced as more personal and intimate than SNSs (Boczkowski, Matassi, & Mitchelstein, 2018), we can expect disconnection to be a necessary coping strategy. We thus ask:

RQ1:

To what extent do IM users engage in disconnective behaviors on IM platforms, including (a) dissociation and (b) content filtration?

RQ2:

Why do IM users engage in disconnective behaviors on IM platforms?

Political disagreement and disconnection

Existing research has mainly focused on political disagreement as an explanation of politically motivated unfriending. Findings consistently show that exposure to political content one disagrees with or encountering people holding opposing ideologies is predictive of a higher likelihood of unfriending (Barnidge, Peacock, Kim, Kim, & Xenos, 2022; Bode, 2016; John & Dvir-Gvirsman, 2015; Skoric, Zhu, Koc-Michalska, Boulianne, & Bimber, 2021; Zhu & Skoric, 2021). Some thus suggest that unfriending constitutes a form of selective avoidance (Zhu & Skoric, 2021; Zhu, Skoric, & Shen, 2017). This is arguably because political disagreement tends to elicit a state of psychological discomfort and defensive reaction (Kim, Jones-Jang, & Kenski, 2022; Neubaum, Cargnino, & Maleszka, 2021). In times of heightened political conflicts, disagreement is likely perceived as norm violation, line-crossing, and hence a marker for disconnection (Jordá, Cañedo, Bene, & Goyanes, 2021; Neubaum, Cargnino, & Maleszka, 2021, 2021; Schwarz & Shani, 2016).

Based on this line of research, we expect disconnective behaviors on IM platforms to be a reaction to political disagreement as well. Content filtration through muting someone or a group is arguably an effective means to hide content from the sources while keeping the ties. Tie dissolution through blocking or deleting someone, leaving a group, or removing someone from a group redraws the network boundary and prevents future encounters. We thus propose the following hypothesis:

H1.

Exposure to political disagreement is positively associated with disconnective behaviors on IM platforms, including (a) dissociation and (b) content filtration.

Safe space through disconnection

While scholars are primarily concerned that politically motivated unfriending may thwart exposure to political disagreement, recent research advances that it is a means to build so-called “safe spaces”. Cutting ties with sources of political disputes can protect people from negative repercussions such as social and state sanctions, which is especially important for minorities (Zhu & Skoric, 2021, 2022). When examining unfriending of Jewish Israelis by Palestinian citizens of Israel, John and Agbarya (2021) argue: “acknowledging structural power differentials between groups of Facebook users requires us to rethink the meaning of unfriending when carried out on someone higher up a social hierarchy” (p.1074). In sum, this strand of research suggests that people disconnect with those holding opposing political views in response to social risks and structural injustice.

Building on this, we further develop the safe space theoretical framework to examine disconnection on IM platforms in the context of Hong Kong politics. Specifically, we focus on two aspects of “safe space”: First, disconnection takes place during everyday political interactions that operate within spaces of social uncertainty – disagreement evokes anxiety about social disapproval, rejection, and other forms of social penalty (Eliasoph, 1998; Neubaum & Krämer, 2018). Such anxiety is heightened in a conflict-avoidant society like Hong Kong where political topics are sensitive and controversial (Zhang, 2022). Thus, we understand disconnection as a strategy to curate a socially safe space – a space that is safe from social sanctions. Second, research shows that in non-democratic countries and among minorities and marginalized groups in democracies, participation risks state and social surveillance, censorship, and retribution, which encourages people to devise tactics of resistance to manage their visibility (A. Lee, 2018, 2022; Van Duyn, 2021). Therefore, the need for safe spaces stems not only from social, but systemic threats as well, especially for people who are disadvantaged, marginalized, or oppressed (Collective, 2014). Accordingly, we focus on two factors to explain why people curate safe spaces through disconnective behaviors, namely fear of social isolation (i.e., fear of being rejected and ostracized by others) and systemic victimhood (i.e., feelings of victimization by an unjust system).

Fear of social isolation

Fear of social isolation refers to “fear – probably developed over the course of evolution – of being rejected by those around” (Noelle-Neumann & Petersen, 2004, p. 349), “an emotional fear of losing social contacts with others, i.e., others might leave us alone and isolated” (Matthes et al., 2012, p. 291). This is consistent with work in social and developmental psychology that the need to belong is a fundamental human trait (Baumeister & Leary, 1995). As threat of isolation is an effective form of social sanction, fear of social isolation is the basis on which conformity and avoidance largely operate (Matthes et al., 2012; Neuwirth, Frederick, & Mayo, 2007; Noelle-Neumann, 1974).

According to the spiral of silence (SOS) theory, holding a minority or unpopular view can induce fear of social isolation, which motivates conformity such as moderating and silencing one’s own opinion (Noelle-Neumann, 1974). While SOS conceptualizes fear of social isolation as a transitory mental condition, others see it as a dispositional trait or a general tendency of individuals that drives specific behaviors in a given situation (Matthes et al., 2012; Neuwirth, Frederick, & Mayo, 2007). Research shows that this socio-psychological trait is positively correlated with individuals’ inclination to self-censor; those who fear social isolation are more likely to monitor their social environments to gauge others’ opinions, and self-censor when disagreement is perceived from a crowd or an individual to avoid being penalized for what they say (Hayes, Matthes, & Eveland, 2013; Matthes et al., 2012). On social media, fear of social isolation not only leads to expression avoidance, but also motivates withdrawal behaviors such as deleting or editing what one has posted before (Chen, 2018).

To summarize, fear of social isolation drives people to regulate their own expressions and behaviors. How will it influence behaviors that dissolve social ties and remove others’ content in the face of disagreement – behaviors that involve regulating others? There is so far little research on this. On the one hand, one can argue that, for those who are sensitive to social disapproval and have a strong need to belong, cutting ties and filtering content because of political disagreement may seem counterintuitive. On the other hand, fear of social isolation also presents conditions that can facilitate disconnection. First, it can motivate people to monitor their social surroundings, which likely expose them to more political disagreement and make information management a necessity (Skoric, Zhu, & Lin, 2018). Second, those with heightened fear of social isolation tend to overestimate the consistency between others’ opinions and their own (Morrison & Matthes, 2011). This imagined homogeneity is likely to collapse during polarizing events, which evokes shock and disappoint and consequently drives people to unfriend others so as to reestablish a sense of homogeneity (Schwarz & Shani, 2016). However, given the little direct research on this, we propose the following research question:

RQ3.

How does fear of social isolation moderate the relationship between exposure to political disagreement and (a) dissociation and (b) content filtration on IM platforms?

Systemic victimhood

Systemic victimhood refers to the kind of victimhood that “manifests in the form of feelings of victimization by systemic power structures” (Armaly & Enders, 2021, p. 19). It is oppressor-oriented in the sense that, across the political spectrum, those high on systemic victimhood tend to feel that the system is rigged to benefit a selected few and the world is out there to get them; they attribute blame to systemic injustices as causing their victim status (Armaly & Enders, 2021).

We expect feelings of systemic victimhood to exacerbate individuals’ tendency to disconnect in reaction to interpersonal political disagreement. First, victimhood is an important identity position that shapes how individuals constitute themselves and others as political actors. It drives self-protection and fuels the fight for recognition and reparation (Armaly & Enders, 2021; Horwitz, 2018). Thus, people with strong feelings of systemic victimhood are likely to react to political disagreement with a defensive mind-set and perceive political disagreement as personal attacks on their already marginalized positions. This can trigger disconnection as a way to protect themselves from the attacks and to protest against the hegemony of an oppressive system. Furthermore, people seeing themselves as victims of an unjust system are found to be more supportive of political correctness than others, seeing it as an important way to limit systemic prejudice against disadvantaged groups (Armaly & Enders, 2021). It is thus likely that they may see opposing views as exclusionary, disparaging, and insulting, and therefore silence them through muting and dissociation. Accordingly, we propose the following hypothesis:

H2.

Systemic victimhood strengthens the positive relationship between exposure to political disagreement and (a) dissociation and (b) content filtration on IM platforms.

The proposed conceptual model is summarized in Figure 1.

Figure 1. Conceptual model.

Method

We conducted a survey in Hong Kong in June 2021 and contracted the research company Dynata to collect the data. To achieve a diverse panel, Dynata recruited participants from various channels, including online advertisements, loyalty program of its partners, member referral, social media, and so on. In order to obtain a sample of which the demographic characteristics match those of the adult population in Hong Kong, participants were stratified based on age, gender, and income status. Participants were offered one U.S. dollar as compensation for completing the survey. In total, we collected valid responses from 1000 Hong Kong residents aged 18 and above, and the demographic information is listed in Table 1. Compared to Hong Kong’s census data,¹ our sample is slightly younger and reported a higher average household income, as the population aged between 18 and 65 had a mean age of 43.36 at the end of 2020 and a median monthly household income of HKD27,400 in 2021. Among the total sample, 946 participants (94.6%) reported to use IM platforms, specifically, WhatsApp and/or Telegram, the most popular IM platforms in Hong Kong at the time of the study². Non-users were excluded from analysis.

Table 1. Sample demographics (N = 1000).

	Min	Max	M(SD)	Median	%
Age	18	65	39.37(12.29)	39
Gender (female)					51.30
Education level	1	6	4.10(1.51)	5
Household monthly income	2	14	9.75 (3.31)	11

Education: 6-point scale (1= Less than high school, 2= High school degree or equivalent, 3= Some college but no degree, 4= Associate degree, 5= Bachelor degree, 6= Graduate degree).

Family monthly income: 14-point scale (1= “Under HKD7,999,” 2 = “HKD8,000–9,999,” 3 = “HKD10,000–14,999,” 4 = “HKD15,000–19,999,” 5 = “HKD20,000–24,999,” 6 = “HKD25,000–29,999,” 7 = “HKD30,000–34,999,” 8 = “HKD35,000–39,999,” 9 = “HKD40,000–44,999,” 10 = “HKD45,000–49,999,” 11 =”HKD 50,000–59,999,” 12 = “HKD60,000–79,999,” 13 =”HKD80,000–99,999, 14 = “HKD100,000 or above”).

To examine the extent to which people disconnect on IM platforms for political reasons (RQ1) and its relationships with political disagreement (H1), fear of social isolation (RQ3), and systemic victimhood (H2), we adopted established scales to measure the related variables. We also designed an open-ended question to answer RQ2 regarding the subjective meanings of disconnection. Please see details of the measures as follows.

Measures

Disconnective behaviors

In operating disconnective behaviors on IM platforms, we distinguished dissociation from content filtration. We measured the two types of disconnection with a series of binary items that captured various means afforded by IM platforms. For dissociation, we asked respondents whether they had (1) deleted someone’s contact information, (2) blocked someone, (3) left a group, and (4) removed someone from a group on WhatsApp or Telegram for reasons related to politics in the past three months (1 = yes, 0 = no, 3 = don’t know). Respondents who answered “don’t know” were treated as missing data, which were omitted from analysis. The items were combined to create an additive index that measured the extent of dissociation (Range: 0 to 4, M = .65, SD = 1.16). Similarly, we measured the extent of content filtration with two items (Range: 0 to 2, M = .46, SD = .73), including (1) muting a group and (2) muting someone.

Subjective meanings of disconnection

Following the above closed questions, we asked respondents an open-ended question “can you please describe why you did it.” In order to understand from a user perspective why they engaged in disconnective political behaviors on IM platforms, we invited them to recall a recent event in answering this question. The answers were written in English or Traditional Chinese. Among the answers in English, on average, responses were 14 words long with a maximum of 36 words. The responses in Chinese on average had 38 characters in length with a maximum of 88 characters. Some typical responses are, for example, “because some people are too extreme and disagreed with me. I wanted to delete them to make my stand clear,” “avoid people getting a wrong idea of my views.”

Exposure to political disagreement

Following Zhu and Skoric (2021), we measured exposure to political disagreement with two items on a 7-point scale (1 = not at all, 7 = very often). We asked respondents how often in the past three months on WhatsApp or Telegram, they (1) encountered opinions related to politics that were different from their own, and (2) found themselves disagreeing with others on political issues. The variable was computed by taking the mean (rs= .82, p < .001; M = 2.59, SD = 1.54).

Fear of social isolation

We used the established scale to measure fear of social isolation as a dispositional trait (Hayes, Matthes, & Eveland, 2013; Matthes et al., 2012). We asked respondents to what extent they agreed or disagreed to five statements on a 7-point scale (1 = strongly disagree, 7 = strongly agree), for example, “it is scary to think about not being invited to social gatherings by people I know,” and “one of the worst things that could happen to me is to be excluded by people I know.” We estimated the mean to measure fear of social isolation (Cronabach’s α = .86, M = 3.98, SD = 1.22).

Systemic victimhood

We adopted Armaly and Enders (2021) scale to measure self-perceived systemic victimhood. On a 7-point scale (1 = strongly disagree, 7 = strongly agree), we asked respondents to what degree they agreed or disagreed on four statements: (1) the system works against people like me; (2) the world is “doing it” to me and there is nothing I can do about it; (3) the world is out to get me; and (4) the system is rigged to benefit a selected few. The variable systemic victimhood was estimated by taking the mean of the items (Cronabach’s α = .83, M = 4.22, SD = 1.16).

Control variables

We controlled for demographic variables including age (range: 18 to 65, M = 39.04, SD = 12.23), gender (51.5% female), education (range: 1 to 6, M = 4.10, SD = 1.51), and family income (range: 1 to 14, M = 9.73, SD = 3.28). We also controlled for political interest, IM use intensity, and political use of IM platforms. Previous research finds them to predict politically motivated unfriending, and they are also likely associated with encountering political disagreement on social media (Goyanes, Borah, & Gil de Zúñiga, 2021; Skoric, Zhu, & Lin, 2018; Yang, Barnidge, & Rojas, 2017; Zhu & Skoric, 2021). We estimated political interest with a single item “how interested are you in what’s going on in government and politics” (1 = not interested at all, 5= extremely interested; M = 3.09, SD = 1.09). IM use intensity was measured by asking respondents how often they used WhatsApp or Telegram (1 = never, 7 = several times a day; M = 4.50, SD = 1.18). Political use of IM platforms was measured with eight items that tapped into reading, sharing, expressing, and discussing politics and persuading and mobilizing others on WhatsApp or Telegram (6-point scale, 1= never, 6 = everyday; Cronabach’s α = .93, M = 2.06, SD = 1.20). Additionally, we also controlled for social use of IM platforms, measured with two items about using WhatsApp or Telegram to stay in touch with families and friends, and coworkers or classmates (6-point scale, 1= never, 6 = everyday; rs = .62, p < .001; M = 4.11, SD = 1.55). Zero-order correlations are listed in Table 2.

Table 2. Zero-order correlations.

	1	2	3	4	5	6	7	8	9	10	11	12
1. Dissociation
2. Content filtration	.62***
3. Political disagreement	.27***	.27***
4. Systemic victimhood	.18***	.15***	.10**
5. Fear of social isolation	.12***	.12***	.22***	.13***
6. Political interest	.07*	.07*	.33***	−.01	.15***
7. IM use intensity	.11**	.15***	.16***	.10**	.07*	.14***
8. Political use of IM	.30***	.26***	.56***	.10**	.21***	.40***	.31***
9. Social use of IM	.01	.01	.14***	−.00	.06^#	.14***	.18***	.22***
10. Age	−.05	.13***	−.04	−.19***	−.16***	−.09**	−.22***	−.14***	−.01
11. Gender	.03	.02	.11**	.08*	−.06^#	.09**	.03	.09**	.03	.14***
12. Education	.06^#	.04	.13***	.07*	.02	.13***	.09**	.11**	.02	−.14***	.09**
13. Income	.028	.02	.12***	−.03	.00	.10**	−.01	.09**	.05	.19***	.11***	.41***

Note: ^#p < .10, * p < .05, ** p < .01, *** p < .001.

Analysis

To answer RQ1, we reported the frequencies and percentages of participants who engaged in disconnective behaviors. To answer RQ2, we performed an inductive thematic analysis to identify the overarching themes. This involved three rounds of coding, including initial open coding that provided descriptive labels to categorize and summarize each answer, a following round of focused coding that merged the open codes into more abstract categories, and theoretical coding that integrated the major categories to form a larger theoretical scheme (Braun & Clarke, 2006). In order to answer RQ3 and test H1and H2, we estimated a series of multivariate OLS regressions and moderation analyses.

Results

The extent of disconnection on IM platforms

To answer RQ1, we find that about 31% of IM users reported to have engaged in at least one form of disconnection for political reasons. In terms of dissociation, 21.9% (N = 202) of respondents reported to have left a group and 16.4% (N = 149) blocked others, followed by 15.4% (N = 141) who reported to have deleted someone’s contact and 12.3% (N = 113) removing someone from a group. In terms of content filtration, 27.5% (N = 253) reported to have muted a group and 18.8% (N = 172) muted someone. In general, content filtration is a more common disconnection tactic than tie dissolution Figure 2.

Figure 2. Frequency of disconnectivity on IM platforms.

Why IM users disconnect

To answer RQ2, we conducted a thematic analysis to understand how and why IM users disconnected in their political encounters. Through three rounds of coding, we found political disagreement as a main trigger for disconnection, and the following three dominant themes emerged as to how people experienced political disagreement and disconnected in response. By political disagreement, many referred to differing political beliefs and values and endorsement of opposing political camps.

Theme 1: disconnect to protect personal well-being

Statements like “I don’t want to see the views I disagree with” came up often in many participants’ descriptions, which seem consistent with the existing conceptualization of disconnection as a form of selective avoidance (Zhu & Skoric, 2021; Zhu, Skoric, & Shen, 2017). However, further analysis shows that disconnection is not only a means to filter out disagreements, but to cleanse what is perceived as contamination of one’s personal space. Some called the politically incongruent messages “noisy”, “annoying”, “vulgar”, “ugly”, or “rubbish”, thus disconnecting with the sources to “clear my eyes” and “keep my mind clean”. Others justified their decisions citing personal taste: “I don’t like their views. They are too narrow-minded”.

Another common response is “I don’t want to argue”. While this indicates that disconnection is a means of conflict avoidance (Neubaum, Cargnino, & Maleszka, 2021), we find that the main consideration here is to protect one’s own well-being rather than social harmony. Specifically, some participants saw political discussion as “endless”, “unconstructive”, and “ineffective” as they could “hardly change others’ views”, and thus a “waste of time”. Others see disconnection as a way for emotional regulation. As some explained: “I don’t want to feel angry for them”. “Arguing over politics makes me unhappy”.

Under this theme, participants resorted to both tie dissolution and content filtration, ranging from muting groups to blocking someone. They do so to cleanse and mold their personal spaces to their liking and regulate their emotions. This corroborates previous findings that unfriending is a means to control one’s personal public sphere (John & Gal, 2018), and follows the logic of self-care (Barnidge, Peacock, Kim, Kim, & Xenos, 2022). More importantly, we find that this is in response to the larger political environment: The looming feeling of powerlessness under the NSL (“nothing I can do about it”) made some retreat and focus on what they could control in their personal spaces. As a participant wrote, “I’m powerless in this large environment … but [in my WhatsApp], I just want to hear voices that share my beliefs, and filter out noises that harass me.”

Theme 2: disconnect to protect personal safety

Many participants reported to feel “unsafe,” “uncomfortable,” or “disturbed” when encountering political disagreement in their IM environments. A sense of distrust and suspicion toward those expressing views that endorsed the opposing political camp ran deep here, which evoked fear for personal safety.

Three sources of fear that prompt disconnection emerged in our analysis. First, some feared social surveillance when seeing opposing views in their IM circles. For example, one participant wrote, “there are too many people in Hong Kong who are CCP [Chinese Communist Party] people, thugs … I’m afraid of being betrayed.” Second, some feared real-life repercussions such as harassment and physical retaliation, as one participant said, “I am worried that my families will be disturbed, harassed, and hurt.” Third, some worried about getting involved in emotionally charged and toxic interactions, as they described them as “irrational,” “unfriendly,” and “disrespectful.”

In response to these fears for personal safety, many participants resorted to disconnection in order to regulate their own visibility. A common sentiment is “I don’t want to be seen” and “I don’t want others to know my political stance,” and they did so primarily through tie dissolution, such as deleting and blocking someone and removing someone from a group.

Theme 3: disconnect to reduce social risks

Another theme that emerged from the participants’ accounts is that they disconnect on IM platforms to minimize social risks. There are two types of social risks, including having problematic social affiliations and jeopardizing otherwise harmonious relationships.

As to the former, many participants reported that what they perceived as the “extremes,” “radicals,” “vulgar words,” or “saying things they should not have” signaled to them a problematic affiliation. This is similar to previous findings that the culprit is moral violation, incivility, and other forms of line-crossing (Schwarz & Shani, 2016). Moreover, in the context of Hong Kong under the NSL, these affiliations are “wrong” and “dangerous” in the sense that they may invite state and social surveillance and retribution. Such social risks prompted people to cut ties completely through blocking contacts and leaving groups.

Others reported to mute groups or someone on IM platforms to avoid confrontation. In doing so, they turned away from political disagreement in order to protect personal relationships. We highlight here that in these cases, participants often reconcile with political disagreement by downplaying politics – years of protests and social movements were to no avail, and now under the NSL political discussion is something “meaningless” and “useless”. For example, a participant described:

[I mute groups to] Avoid quarrelling with others and violating the national security law or related regulations, and try to maintain harmony among the circle of friends. While political views may differ from person to person, it is not worth fighting for it, nor will it promote socio-political progress.

Overall, the thematic analysis shows that participants disconnect to curate safe personal spaces in response to an uncertain and insecure political environment. Specifically, they sever ties with those holding questionable views to regulate their own visibility and remove problematic affiliations to avoid being targeted by authorities and peers. They filter content to protect personal relationships and well-being from contentious political discourse, largely because they feel powerless in affecting political change.

Predict disconnective behaviors

To test H1 regarding the relationship between political disagreement and IM disconnective behaviors, we performed two OLS regressions predicting dissociation and content filtration respectively (see Table 3). Results show that while controlling for other variables, political disagreement is positively associated with both dissociation (b(SE) = .12(.03), p < .001) and content filtration (b(SE) = .09(.02), p < .001). H1 is thus supported.

Table 3. Summary of OLS regressions predicting disconnectivity on IM platforms.

	Dissociation		Content filtration
	b(SE)	β	b(SE)	β
Political disagreement	.12(.03)***	.16***	.09(.02)***	.19***
Systemic victimhood	.14(.03)***	.14***	.06(.02)**	.09**
Fear of social isolation	.03(.03)	.04	.02(.02)	.04
Age	.00(.00)	.02	−.01(.00)*	−.08*
Gender (0= female, 1= male)	−.01(.08)	−.01	−.01(.05)	−.01
Education	.02(.03)	.02	−.02(.02)	−.03
Income	−.00(.01)	−.01	.01(.01)	.03
Political interest	−.08(.04)*	−.07*	−.05(.02)^#	−.07^#
IM use intensity	.02(.03)	.02	.05(.02)*	.07*
Political use of IM	.22(.04)***	.23***	.09(.03)**	.14**
Social use of IM	−.04(.03)^#	−.06^#	−.03(.02)	−.05
Constant	−.65(.32)*		−.04(.20)
Ajusted R^2	.13		.11
F(df)	12.66(11, 874)***		11.31(11, 892)***

Note: ^#p < .10, * p < .05, ** p < .01, *** p < .001.

To test whether and how fear of social isolation and systemic victimhood moderate the relationship between political disagreement and disconnection (RQ3 and H2), we entered two interaction terms to the OLS regression models (see Table 4). The independent variables are mean centered to avoid multicollinearity issues. We find that although fear of social isolation is not directly predictive of disconnection, it strengthens the positive relationships between political disagreement and tie dissociation (b(SE) = .06(.02), p < .01) and content filtration (b(SE) = .03(.01), p < .05). Systemic victimhood strengthens the relationship between disagreement and tie dissociation (b(SE) = .04(.02), p < .05), but not content filtration (b(SE) = .01(.01), p = .34). H2a is supported but H2b is not.

Table 4. Summary of OLS regressions predicting disconnectivity on IM platforms with interaction effects.

	Dissociation	Content filtration
	b(SE)	b(SE)
Political disagreement	.10(.03)***	.08(.02)***
Systemic victimhood	.14(.03)***	.06(.02)**
Political disagreement × systemic victimhood	.04(.02)*	.01(.01)
Fear of social isolation	.04(.03)	.02(.02)
Political disagreement × fear of social isolation	.06(.02)**	.03(.01)*
Age	.00(.00)	−.01(.00)*
Gender (0 = female, 1= male)	−.00(.07)	−.01(.05)
Education	.02(.03)	−.01(.02)
Income	−.01(.01)	.01(.01)
Political interest	−.07(.04)^#	−.04(.02)^#
IM use intensity	.03(.03)	.05(.02)*
Political use of IM	.21(.04)***	.08(.03)**
Social use of IM	−.04(.02)^#	−.02(.02)
Constant	.38(.27)	.52(.17)**
Ajusted R^2	.15	.13
F(df)	12.14(13, 872)***	10.23(13, 890)***

Note: ^# p < .10, * p < .05, ** p < .01, *** p < .001.

Discussion and conclusion

Weighing in on the existing literature that the closed digital spaces of IM platforms can facilitate a discursive democracy, this study examines how users actively curate these closed digital spaces, with a focus on disconnective behaviors including tie dissolution and content filtration. We conduct the study in the context of Hong Kong politics one year after the implementation of the National Security Law. Combining statistical analysis of survey data and thematic analysis of open-ended survey questions, we estimate the extent to which IM users engage in disconnective behaviors, understand the associated meanings, and explain the behaviors with a focus on the socio-psychological and structural conditions, namely fear of social isolation and perceived systemic victimhood. This study makes the following contributions.

First of all, throughout the thematic and statistical analyses, political disagreement emerged as a main factor that led to disconnection on IM platforms, including both dissociation and content filtration. While this is consistent with the existing research on politically motivated unfriending, we move the field forward by demonstrating that disconnection is a product of systemic problems, rather than individual choices alone. Specifically, our statistical analysis shows that political disagreement is a stronger predictor of tie dissolution among those seeing themselves as victims of an oppressive system. They are more likely to delete or block someone, leave a group, and/or remove someone from a group when encountering counter-attitudinal views. With the findings from the thematic analysis, we locate this sense of systemic victimhood within the context of Hong Kong politics: First, many see those holding opposing political views with suspicion and cut ties with them in order to regulate their own visibility, fearing that those from the other side would monitor, confront, harass, and report them. Second, some cut ties with their IM contacts that appeared to be politically incorrect, extreme, or radical, worried that keeping these problematic contacts runs the risk of violating the NSL. Third, some feel powerless that discussing politics on IM platforms is not only dangerous but also meaningless and ineffective, and thus disconnect to cleanse their personal spaces of disagreeing voices. In sum, these findings suggest that those feeling greater systemic threats tend to cut ties to protect their personal safety.

Moreover, individuals’ tendency to disconnect when facing political disagreement is also dependent on their socio-psychological tendencies. Specifically, we find that fear of social isolation strengthens the positive relationship between political disagreement and content filtration. This suggests that when confronted with opposing political views, those with strong fear of social isolation tend to mute contacts and groups (while keeping the connections). According to the thematic analysis results, content filtration here is a tactic to turn away from contentious political discourse – to exercise self-care, avoid conflict, and protect personal relationships. First of all, this is consistent with the literature on informal political talk that politics can evoke social anxiety and fear for jeopardizing relationships (Eliasoph, 1998). More importantly, we highlight that such social anxiety and fear should be understood within a given political environment: As the thematic analysis suggests, many downplayed the importance of politics because of the prevailing sentiment of powerlessness and danger under the NSL, which prompted them to mute political content and instead focus on personal relationships. We can make sense of this finding with the Terror Management Theory that existential threat can encourage relationship building as a buffering mechanism (Pyszczynski, Solomon, & Greenberg, 2015). This may also explain why systematic victimhood does not moderate the relationship between political disagreement and content filtration: Content filtration is a disconnection tactic mainly aimed at protecting social relations and well-being from contentious political discourse, which is arguably a more universal need among the population.

Lastly, we find fear of social isolation to strengthen the positive relationship between political disagreement and tie dissolution on IM platforms. This means that those fearing rejection or ostracism are more likely to delete/block others, leave a group, and/or remove someone from a group because of political disagreement. While this may seem counterintuitive, it is arguably because these people tend to overestimate the political similarity between others’ opinions and their own (Morrison & Matthes, 2011), which can lead to shock and disappointment when discovering different and sometimes unacceptable views from people they consider friends (John & Agbarya, 2021; Schwarz & Shani, 2016). Moreover, considering the Hong Kong political context, they may cut ties with those expressing views that are problematic under the NSL, fearing that keeping those contacts will lead to social disapproval from other people.

This study is to our knowledge the first to examine disconnective behaviors on IM platforms, which contributes to the literature of politically motivated unfriending, political talk, and safe spaces in everyday political interactions. It benefits from a mixed-method approach combining statistical analysis of survey data and thematic analysis of open-ended questions. The quantitative findings identify the extent and conditions under which individuals resort to disconnection in curating safe spaces in their IM environments, and the qualitative findings contextualize these patterns within the given political context and from the users’ perspective. Despite this strength, the study does have limitations. First, our data is collected through a cross-sectional survey, which tends to be weak in inferring causal relations. The two moderators – systemic victimhood and fear of social isolation – are relatively stable tendencies or traits. In comparison, the causal link between political disagreement and disconnection is harder to establish, although it is corroborated by the findings from the thematic analysis. Second, the qualitative data collected through open-ended questions is limited in capturing individuals’ lived experiences. We thus recommend future research to employ in-depth interviews to offer finer-grained understandings. Third, this study is conducted within the context of Hong Kong after the introduction of the National Security Law. It is thus important to contextualize the findings in a non-democratic political environment with growing state and social sanctions. At the same time, state and social sanctions are not unique to non-democratic regimes, and systemic threats are real to minorities and marginalized groups in democracies. We thus urge future research to examine the topic in other political contexts and through cross-country comparative designs.

To conclude, contrary to the common belief that IM platforms offer controlled, closed, and hence safe social settings, this study highlights that users actively curate such safe spaces through tie dissolution and content filtration in response to what they perceive as unsafe in a given political and social environment. Existing literature on politically motivated unfriending understands disconnection as a form of selective avoidance, and presents a worrying view that it leads to echo chambers and prevents people from hearing the other side – a conclusion that is drawn mostly from liberal democracies and based on the normative theory of deliberation. This study highlights that disconnection is not simply a tool to shelter from disagreement, but a product of structural issues and a necessary response to social and systemic threats that come with political disagreement: Disconnection creates safe spaces of withdrawal following the logic of self-care, where people turn away from contentious political discourse perceived as dangerous, meaningless, and ineffective, and focus instead on the protection of their own personal safety, well-being, and relationships.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

The work was supported by the City University of Hong Kong

Notes on contributors

Qinfeng Zhu

Qinfeng Zhu is an Assistant Professor at the Centre for Media and Journalism Studies at the University of Groningen, the Netherlands. She holds a Ph.D. in Communication from the City University of Hong Kong and a M.A. in Communication from the National University of Singapore. Her research examines how emerging technologies shape citizens’ social relations and information environments, and its implications for citizens’ political and civic (dis)engagement.

Marko M. Skoric

Marko M. Skoric is an Associate Professor at the Department of Media and Communication, City University of Hong Kong. He holds a Ph.D. in Communication from the University of Michigan, and a B.Sc. in Psychology from the University College London, UK. Marko’s teaching and research interests are focused on new media and social change, with a particular emphasis on the civic and political implications of new communication technologies.

Notes

1. Census and Statistics Department, The Government of the Hong Kong Special Administrative Region: https://www.censtatd.gov.hk/en/scode150.html

2. Top apps ranking: https://www.similarweb.com/apps/top/google/store-rank/hk/communication/top-free/