Introduction
Since the launch of platforms such as MySpace and Facebook in the early 2000s, social media usage has expanded rapidly, with over 5 billion global users by 2024 (Kemp, Reference Kemp2024). This expansion has been driven by increased internet access and the proliferation of low-cost smartphones, allowing individuals in various political settings to engage with social media platforms (Silver et al., Reference Silver, Huang and Taylor2019). In hybrid and authoritarian regimes, this raises important questions about how social media use relates to citizen attitudes toward civil society and state institutions. While social media may open space for new forms of political engagement, governments have adapted by seeking to regulate digital spheres or by deploying these platforms as tools of state messaging (Deibert, Reference Deibert2019).
Under authoritarian and hybrid regimes, limited “venting” outlets and constrained freedoms of expression coexist with strict suppression of dissent. While regimes tolerate some apolitical discourse, grievances related to restricted liberties or pro-democracy advocacy are dismissed. In this context, social media platforms have emerged as spaces for circumventing state-controlled narratives and growing civic engagement through several mechanisms. Social media enables access to information and perspectives absent from traditional media channels, which are often censored or co-opted by regimes. By bypassing gatekeepers, users gain exposure to independent reporting, transnational activism, and counter-narratives that challenge regime propaganda. Civil society organizations, particularly nongovernmental organizations (NGOs), leverage social media strategically to expand their missions and build networks of solidarity. These platforms allow NGOs to employ targeted campaigns and engage citizens through online interactions.
These mechanisms strengthen trust in NGOs by connecting individuals to advocacy groups and creating collective identities (Bennett and Segerberg, Reference Bennett and Segerberg2012). For instance, social media’s role in disseminating uncensored information empowers citizens to critically evaluate regime claims, while NGOs help them translate public grievances into demands for accountability. Over time, such dynamics can erode authoritarian legitimacy and expand the reach of civil society, with broader implications for democratization and institutional effectiveness (Herrold and AbouAssi, Reference Herrold and AbouAssi2023). At the same time, the effects of social media use on trust in government institutions remain contested. One line of argument, rooted in Putnam’s (2000) theory of social capital, suggests that increased interpersonal connectivity may generate generalized trust, including toward governance institutions. From this view, social media could bolster trust in governments that support pluralism and demonstrate accountability. In contrast, digital spaces may amplify public grievances, increase exposure to state failings, and facilitate citizen mobilization against incumbent regimes (Bellin, Reference Bellin2012; Tudoroiu, Reference Tudoroiu2014). In authoritarian settings, where transparency is limited, social media can circulate reports of corruption and repression, undermining public confidence in the state (Reuter and Szakonyi, Reference Reuter and Szakonyi2015; Enikolopov et al., Reference Enikolopov, Petrova and Sonin2018). Many governments have responded by attempting to curtail civil society and regulate online spaces (King et al., Reference King, Pan and Roberts2013; Dupuy et al., Reference Dupuy, Ron and Prakash2014; Ong, Reference Ong2021; Chaudhry, Reference Chaudhry2022).
This paper explores how social media use relates to trust in civil society and government institutions across Southeast Asia. This is a region where internet penetration is high, and political systems range from flawed democracies to consolidated authoritarian regimes. Countries such as Malaysia, the Philippines, and Thailand rank among the most active globally in daily internet usage (Kemp, Reference Kemp2024), and many have introduced restrictions on NGOs and social media platforms while facing sustained levels of online political participation (Abbott, Reference Abbott2011; Sinpeng, Reference Sinpeng2021). Using the Asian Barometer, I examine how social media usage and expression of political opinions online is associated with trust in civil society organizations and governance institutions. The analysis finds that individuals who use social media to express their political opinions report higher levels of trust in NGOs, but lower levels of trust in a range of governance institutions, from the executive to the local level. These findings contribute to our understanding of the complex relationship between social media engagement and institutional trust in Southeast Asia.
Literature
Trust and media
Trust is a reciprocal relationship where actors feel that cooperation and inputs into a relationship will be mutually respected. As Kydd (Reference Kydd2000) notes, interpersonal trust means that actors believe that the other side is likely to reciprocate cooperation rather than opportunistically exploit it. There is a relational aspect to trust whereby individuals make themselves vulnerable to others (Levi and Stoker, Reference Levi and Stoker2000). Institutions are part of these trust relationships. One important component of institutional efficiency is trust or perceptions of trustworthiness from the users of said institutions. Institutions with high levels of trustworthiness can induce higher levels of political participation and even shifts in perceptions on democracy (Sztompka, Reference Sztompka1999; Inglehart, Reference Inglehart1990). Institutional trust can also be understood as a component of political legitimacy (Tyler and Jackson, Reference Tyler and Jackson2014). Furthermore, higher trust in institutions is often associated with more accountable and accessible governments, high social capital, system support, and better social cohesion (Putnam, Reference Putnam2000; Paxton, Reference Paxton2002; Yap, Reference Yap2019). Determining where trust in institutions lies is an important aspect of regime development and democratic growth.
Trust in institutions is enhanced through open communication between governors and the governed. Traditionally, newspapers and television served as primary political communicators. New media forms have reshaped civic spaces, with the internet and smartphones enabling over 63% of the global population to use social media (Shewale, Reference Shewale2024). Bennett (20, Reference Bennett2003) argues that social media platforms “overcome limits in time, space, identity, and ideology,” expanding political coordination. This shift reflects the mechanisms behind social media’s role in democratizing access to uncensored information and enabling civil society to connect with the public strategically.
Questions remain over whether social media translated to real-life civic participation and thus greater interaction with civic institutions. Studies have linked social media use with higher participation in political behavior and civic participation (Bode, Reference Bode2012; Xenos et al., Reference Xenos, Vromen and Loader2014; Boulianne, Reference Boulianne2015; Gainous et al., Reference Gainous, Abbott and Wagner2019). However, with increased online civic participation comes downsides. Valenzuela et al. (Reference Valenzuela, Halpern, Katz and Miranda2019) find that in Chile increased social media usage for news was often associated with the spreading of misinformation. Other studies have found that disinformation and misinformation is mostly spread by individual citizens on social media (Golovchenko et al., Reference Golovchenko, Hartmann and Adler-Nissen2018; Allcott et al., Reference Allcott, Gentzkow and Yu2019). Therefore, despite the increased civic participation of individuals, social media can bring out an element of distrust and uncertainty. Social media users with lower levels of political knowledge are more likely to distrust news media (Stubenvoll et al., Reference Stubenvoll, Heiss and Matthes2021). Furthermore, social media also shapes trust in institutions on partisan lines, leading to polarized trust in the government (Klein and Robison, Reference Klein and Robison2019).
The literature on trust and social media has mostly focused on Western, democratic states. Studies on trust in non-Western, hybrid, and authoritarian states have focused on the paradox of high trust in governments despite repression (Zhai, Reference Zhai2018; Nathan, Reference Nathan2020). However, as citizens become more educated and more exposed to social media discourses, do their levels of trust change? Nathan (165, Reference Nathan2020) states “the more authoritarian regimes succeed in their modernizing mission, the more they undermine the prevalence of traditional social values among their populations.” With the spread of liberal democratic values through education and global internet access, the inevitable is possible. Some predict the erosion of this paradox of trust over time due to concerns with regime effectiveness and exposure to the internet (Zhong and Zhan, Reference Zhong and Zhan2021; Chen, Reference Chen2017).
Under competitive authoritarianism, social media enables limited critique and civil community formation. Access to dissident information on social media can increase negative government perceptions, while strategic digital networks connect users to mobilize protests (Gainous et al., Reference Gainous, Wagner and Abbott2015, Reference Gainous, Abbott and Wagner2019). However, regimes can also restrict platforms and co-opt messaging (Howard et al., Reference Howard, Agarwal and Hussain2011; Gunitsky, Reference Gunitsky2015). These dynamics suggest social media use inversely correlates with trust in hybrid and authoritarian governance.
Creating civic spaces: NGOs and social media
Nonprofit organizations and NGOs play a significant role in service provision and advocacy. They are often perceived as principled actors distinct from firms, but they are also strategic entities concerned with their operations and funding (Keck and Sikkink, Reference Keck and Sikkink1998; Johnson and Prakash, Reference Johnson and Prakash2007; Sell and Prakash, Reference Sell and Prakash2004). The same tools that a firm would use to advertise itself and communicate with customers, an NGO would also find useful for its purposes. Social media has become an important site for the engagement of volunteers and donors by civil society organizations. Increased usage by NGOs of social media as a space for campaigning, soliciting donations, and promoting certain causes exposes them to more social media users.
However, this exposure can have a dark side. High-profile scandals such as those involving Oxfam in Haiti, WWF in Congo, and Amnesty International, are more likely to be encountered with internet access. Goncharenko (Reference Goncharenko2023) finds that the emergence of the #MeToo movement added to the visibility of nonprofit scandals on social media. The apologies offered by organizations like Oxfam International were deemed inadequate and pointed to a general problem global NGOs face with crisis communication in the social media age (Janssen Danyi, Reference Janssen Danyi2020). NGOs use social media in a stilted and unidirectional fashion, often not engaging with users on platforms and instead emphasizing messaging without interaction due to fear of loss of narrative control (Quinton and Fennemore, Reference Quinton and Fennemore2013; Comfort and Hester, Reference Comfort and Hester2019).
Beyond reputational management, NGOs face structural challenges, such as the growth of “briefcase” NGOs which siphon foreign funding, crowd out local organizations and diminish NGO legitimacy (Prakash and Potoski, Reference Prakash and Potoski2016). Governments, aware of external influence, impose funding restrictions that further constrain NGO operations (Bloodgood et al., Reference Bloodgood, Tremblay-Boire and Prakash2013; Dupuy et al., Reference Dupuy, Ron and Prakash2016). With intense competition for resources, NGOs are therefore highly mindful of their reputations and thus their image as trustworthy actors. If the public are distrustful of NGOs, then they will operate in hostile environments without local support. This is a vicious cycle which then further entrenches their reputation as foreign actors especially since they lack incentives to push for locally specific goals (Burger and Owens, Reference Burger and Owens2010). The visibility of negative discourse on social media may further erode public trust in NGOs.
However, despite these issues with civil society, the influence of social media can often lead to positive engagement. Although NGOs often use social media in a “one-way” fashion (Lovejoy et al., Reference Lovejoy, Waters and Saxton2012; Guo and Saxton, Reference Guo and Saxton2013), these studies have often focused on organizations in the Global North. Research on consolidating democracies finds that NGOs often use social media in an interactive way and seek engagement with potential donors and volunteers (Adjei et al., Reference Adjei, Annor-Frempong and Bosompem2016; Armstrong and Butcher, Reference Armstrong and Butcher2018). Furthermore, social media usage can enhance the image of an NGO and promote trust in its brand (Gregory et al., Reference Gregory, Ngo and Miller2020, Bilgin and Kethüda, Reference Bilgin and Kethüda2022). Social media can also be a powerful space for civic engagement and NGOs can speak directly to users without direct state intervention (Leggett, Reference Leggett2017; Vu et al., Reference Vu, Blomberg, Seo, Liu, Shayesteh and Do2020).
In authoritarian contexts, NGOs adapt their strategies to navigate internet restrictions. NGOs on Weibo (a popular social media platform in China) use colloquial language and emojis to communicate with followers and build legitimacy (Zhou and Pan, Reference Zhou and Pan2016). These interactions can build legitimacy and create community spaces online despite internet restrictions. Independent media is also able to engage with users more freely on social media platforms, especially in repressive states where news broadcasts might be more tightly controlled. Lee et al. (Reference Lee, Johnson and Prakash2010) found that in the former Soviet Union independent media is positively associated with trust in NGOs. Therefore, social media engagement can potentially enhance trust, provide opportunities for engagement, and legitimacy-building. Furthermore, despite issues facing NGOs, Chapman et al. (Reference Chapman, Hornsey and Gillespie2020) found that global trust has increased in the nonprofit sector.
Southeast Asia case
The Southeast Asia region is characterized by emerging democracies, competitive authoritarian and authoritarian regimes, and a resilient civil society. There is a high level of mobile phone and internet penetration in the region. Over 400 million people use the internet and internet penetration is over 70% in most countries (Von Kameke, Reference von Kameke2024). Most Southeast Asians access the internet via mobile phones (over 70%), which shapes how users interact with online content (Sinpeng and Tapsell, Reference Sinpeng and Tapsell2020). Instead of long-form articles and websites, usage is often centered on shorter videos distributed via platforms such as WhatsApp and TikTok. While most internet users are under the age of 35, older users are increasingly engaging with mobile-based platforms, which has led to concerns from younger users about the spread of misinformation in group chats dominated by older participants (ibid).
Government agencies and public officials frequently communicate through social media, and political discourse is vibrant on these platforms. The electoral success of the Move Forward Party in Thailand has been partly attributed to its strategic use of social media (Habib and Habib, Reference Habib and Habib2023). In Indonesia’s 2024 election, TikTok was critical in shaping political engagement, with Prabowo Subianto’s campaign portraying him as a “gemoy” (cute and lovable) figure to appeal to younger voters (Wahid, Reference Wahid2024). At the same time, states deploy social media to advance repressive aims. The Myanmar military, for example, has used Facebook as part of a broader misinformation campaign targeting the Rohingya population (Kyaw, Reference Kyaw2021).
Given the region’s young population and growing access to smartphones, it is essential to assess how exposure to these platforms has affected institutional trust. Existing research shows that trust in governmental institutions remains relatively high across many Asian countries (Huang and Schuler, Reference Huang and Schuler2018; Jiang and Zhang, Reference Jiang and Zhang2021). In highly centralized political systems such as China, trust tends to be hierarchical, with citizens placing more confidence in central government authorities than in local officials (Zhai, Reference Zhai2022). This pattern also appears in Southeast Asia’s authoritarian regimes, such as Vietnam, where reported trust in institutions remains strong (Ikeda, Reference Ikeda2013). Some scholars argue that such trust may reflect the influence of so-called “Asian values,” which emphasize paternalism and harmony as key political virtues (Wong et al., Reference Wong, Wan and Hsiao2011). Others suggest that this trust may be shaped by social desirability bias or by how people understand the concept of trust itself, often relating it to competence or quality of life under the regime (Chu and Huang, Reference Chu and Huang2010; Cheema and Popovski, Reference Cheema and Popovski2013).
Nevertheless, social media introduces new dynamics that may challenge the persistence of such trust. By facilitating exposure to dissenting viewpoints and revealing state failures or contradictions in official narratives, social media can erode confidence in public institutions. In contexts where governments maintain tight control over information, increased access to these platforms may undermine institutional trust over time, particularly among younger, more digitally engaged populations.
The effects of social media on trust in civil society, especially NGOs, are more ambiguous. The region has a growing NGO sector, though much of the literature has focused on Western and East Asian contexts. Civil society in Southeast Asia operates under a wide array of legal and political constraints (Springman et al., Reference Springman, Malesky, Right and Wibbels2022), but the number of NGOs has grown steadily since political liberalization began in the 1980s (Rodan, Reference Rodan2022). Many of these organizations drew inspiration from Western social movements and focus on service provision or advocacy (Loh, Reference Loh2004). In contrast to the state’s tightly managed messaging, NGOs frequently use social media to engage the public, coordinate action, and advocate for policy change (Lim, Reference Lim2023). In this sense, social media serves as an important tool for circumventing official barriers and amplifying the voices of civil society actors. Public perceptions are shaped by a combination of cultural norms, colonial legacies, and traditions of centralized authority. Some NGOs have gained public trust by addressing local needs and encouraging civic engagement, even within restrictive political contexts (Brass, Reference Brass2022). In these settings, NGOs are not only seen as service providers but also as potential channels for political participation, especially for individuals who lack access to formal political institutions. Publics perceive NGOs as spaces where they can safely express concerns, support causes, and engage in collective action without directly confronting state power (Prakash and Gugerty, Reference Prakash and Gugerty2010; Hansson and Weiss, Reference Hansson and Weiss2024). However, while many NGOs are respected for delivering essential services, their involvement in politically sensitive issues can generate skepticism, particularly in environments where dissent is discouraged (Heurlin, Reference Heurlin2010).
Social media amplifies these tensions. On the one hand, it offers NGOs direct channels to communicate with citizens and mobilize support. On the other hand, it exposes them to accusations of elite bias or foreign influence, particularly in polarized or authoritarian contexts (Lim, Reference Lim2023). Even so, social media may ultimately increase trust in NGOs by creating online civic spaces where citizens can participate in horizontal forms of engagement and activism. While exposure to social media may erode confidence in state institutions, it could simultaneously contribute to the growth of civil society by aiding community-building, peer-to-peer dialog, and new forms of political participation.
Hypotheses
For this study, the relationships between trust and social media are formalized with the following hypotheses:
H1: Individuals who use social media are more likely to report lower levels of trust in political institutions.
H1a: Individuals who use social media to express political views are more likely to report lower levels of trust in political institutions.
H2: Individuals who use social media are more likely to report lower levels of trust in protective institutions, due to their perceived alignment with the state.
H2a: Individuals who use social media to express political views are more likely to report lower levels of trust in protective institutions.
H3: Individuals who use social media are more likely to report higher levels of trust in NGOs.
H3a: Individuals who use social media to express political views are more likely to report higher levels of trust in NGOs.
Political institutions refer to bodies of direct governance, where citizens either elect representatives or, in more restrictive regimes, demonstrate support for ruling parties through controlled electoral processes. In this study, political institutions include national and executive bodies, local government, political parties, parliaments, and courts. Courts are included because, despite their formal neutrality, they often function as political actors in many of the countries examined. Protective institutions represent indirect governance, although citizens do not elect officials in these bodies, such as the police or military, these institutions are closely tied to the functioning of the state. Because protective institutions enforce state authority in everyday life, citizens may associate them with the regime’s legitimacy and accountability. Trust in both political and protective institutions reflects broader attitudes toward state authority, even if the institutions differ in their degree of democratic accountability.
Social media complicates these trust relationships. General social media use, such as consuming news, entertainment, or connecting with others, may expose users to a wide range of political content, including critiques of elite corruption or policy failures. Yet such exposure may not always provoke a strong reaction, particularly in settings where public dissent carries social or legal risk. In contrast, the active expression of political views online, such as posting opinions, sharing political content, or publicly criticizing public officials, signals not only engagement but also a willingness to challenge or affirm political authority in a public space. This form of expression reflects a deeper level of politicization and may correspond with greater scrutiny of state institutions. Accordingly, the expression of political opinions online is more likely than general use to be associated with declining trust in political and protective institutions, especially under regimes that restrict dissent. At the same time, political expression may also reinforce trust in NGOs and other civil society actors that offer alternative forms of accountability and advocacy. By distinguishing between general use and expressive engagement, this study assesses whether political expression online reshapes institutional trust across different levels of governance.
Data and methods
The Asian Barometer Survey (ABS) has been monitoring trust in institutions since the early 2000s. The survey is conducted by the Hu Fu Center for East Asia Democratic Studies at National Taiwan University, and it is the most comprehensive survey of the Asian region on a variety of social and political topics. Respondents are polled on a range of questions which measure political beliefs, attitudes on traditionalism, economic perceptions, and demographic data. For this paper, I am using both Wave 4 and Wave 5 of the ABS. Wave 4 was collected in 2013–2016 and Wave 5 was collected in 2018–2021 across Asia. For the purposes of this study, I have only included the Southeast Asian region. In Wave 4 these countries are: the Philippines, Thailand, Indonesia, Malaysia, Singapore, Myanmar, Vietnam, and Cambodia. In Wave 5 these countries are: the Philippines, Thailand, Indonesia, Vietnam, Malaysia, and Myanmar. The data for Wave 5 is not available for certain countries that were included in Wave 4 (Singapore and Cambodia).
The Asian Barometer includes a set of questions on political trust, asking respondents to rate their trust in various institutions on a four-point scale ranging from 1 (“a great deal of trust”) to 4 (“do not trust at all”). This study draws on items from questions Q7 to Q19, covering the national government, parliament, political parties, local government, the executive, courts, the police, the military, and NGOs. The political items were grouped into two additive indexes: political trust (executive, national government, parliament, political parties, local government, courts), and protective trust (military and police). Each index was reverse coded so that higher values reflect greater trust and rescaled for comparability. Internal consistency of the indexes was assessed using Cronbach’s alpha, the political trust index demonstrates high reliability (α = 0.88), while the Protective Trust Index indicates acceptable reliability (α = 0.77).
Social media engagement is measured using two binary variables. The first captures general social media use based on whether respondents report using any of the three most popular platforms in their country (Q50; 1 = yes, 0 = no). The second measures political expression online, based on whether respondents report expressing their views about politics through social media (1 = yes, 0 = no). Individual-level controls include gender (0 = male, 1 = female), age (continuous), urban residence (0 = rural, 1 = urban), education (three-category ordinal), employment status (0 = unemployed, 1 = employed), and income level (five-category ordinal). Missing data for all trust variables, social media engagement, and controls were addressed using multiple imputation by chained equations. Forty imputations were conducted to reduce sampling variability. The indexes were constructed after imputation, to ensure that missing data did not systematically bias the composition of each trust dimension. Summary statistics for all variables appear in the Appendix (Table 1).
Table 1. Linear regression models predicting political and protective institutional trust
p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001.
To evaluate the hypotheses, I estimate two types of models depending on the outcome of interest, linear regressions with clustered standard errors for trust indices, and ordinal logistic regression for the trust in NGOs outcome. Missing values on key variables were addressed using multiple imputation with predictive mean matching (40 imputations).
For the political and protective trust indices, I estimate the following linear model:
where
-
Y ij is the trust index score for individual i in country j.
-
SocialMedia ij is an indicator of general social media use.
-
Express ij is an indicator of political expression online.
-
γj are country fixed effects.
-
δt are survey wave fixed effects.
-
εij is the individual-level error term.
Tests revealed evidence of heteroskedasticity, so robust standard errors are clustered at the country level to ensure valid inference. Two separate models are estimated: one for political trust (Model 1) and one for protective trust (Model 2). Multicollinearity was assessed using variance inflation factors. All values fall below conventional thresholds, indicating that collinearity is not a concern.
To analyze trust in NGOs, which is measured on a four-point ordinal scale, I estimate a cumulative logit model with the same set of covariates:
$$\eqalign{log[P\left( {{Y_{{\rm{ij}}}}\; \le \;k} \right)\;/\;P({Y_{{\rm{ij}}}}\; \gt \;k)]\;\hskip8pc \cr = \;{\tau _k}\; - \;\left( {{\beta _1}\,SocialMedi{a_{{\rm{ij}}}}\; + \;{\beta _2}Expres{s_{{\rm{ij}}}}\; + \; \cdots \; + \;{\beta _8}Urba{n_{{\rm{ij}}}}\; + \;{\gamma _{\rm{j}}}\; + \;{\delta _{\rm{t}}}} \right)}$$
where
${\tau _{\rm{k}}}$
are threshold parameters for the ordinal outcome. This model (Model 3) also uses clustered robust standard errors by country. The proportional odds assumption was tested and not found to be violated. In addition to the index-based models, I estimate separate ordinal logistic regressions for each individual trust item. These models include the same covariates and appear in the Appendix (Table 2). The results are substantively consistent with the index models.
Table 2. Ordinal logistic regression model predicting trust in NGOs. Estimates from cumulative logit (proportional odds) model
p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001.
Results
Results from models
Table 1 presents results from two models capturing trust in political institutions and protective institutions. In both models, robust standard errors clustered at the country level are used to account for heteroskedasticity and unobserved country-level variation. All models include country and wave fixed effects, and are visualized in Figures 1 and 2. The first model assesses H1, which posits that general social media usage is associated with lower trust in political institutions. The coefficient is small and indistinguishable from zero (0.005, [−0.382, 0.393]), providing no support for H1. In other words, mere use of social media does not predict lower trust in formal political institutions. However, H1a predicts that political expression online is linked to reduced trust. Model 1 indicates that individuals who express their political opinions online score significantly lower on the index of trust in political institutions (−0.47, [−0.71, −0.23]). This finding suggests that expressive engagement with political content online undermines trust in political institutions, consistent with arguments about online political discourse functioning as a venting or critical monitoring mechanism.
Figure 1. Coefficient estimates from linear regression model predicting Political Trust. Each point represents the estimated effect of a variable, with horizontal bars indicating 95% confidence intervals.
Figure 2. Coefficient estimates from linear regression model predicting Protective Trust. Each point represents the estimated effect of a variable, with horizontal bars indicating 95% confidence intervals.
The second model tests H2, which expects that general social media use predicts lower trust in protective institutions, due to their perceived alignment with state power. As with political institutions, the coefficient for social media use is statistically null, offering no support for H2 (−0.015, [0.138, 0.109]). H2a, by contrast, posits that political expression online reduces trust in protective institutions. While the coefficient on expressive behavior is negative, it is not statistically significant (−0.084, [−0.218, 0.050]). Therefore, the results provide partial but inconclusive support for H2a. Taken together, these findings indicate that political expression, rather than exposure, may erode trust in state-aligned institutions, but the effect is more robust for formal political institutions than protective ones.
Model 3 (Table 2, Figure 3) examines trust in NGOs using an ordinal logistic regression. H3 posits that general social media use predicts higher trust in NGOs, while H3a expects that political expression online has a similar effect. The results provide no support for H3, general social media use is not a statistically significant predictor (0.036, [0.080, 0.152]). By contrast, H3a is supported. Individuals who express political views online have significantly higher odds of reporting greater trust in NGOs (1.135, [1.031, 1.251]). This corresponds to a 13.5% increase in the odds of being in a higher trust category for those who engage in expressive political behavior online. This outcome is consistent with the idea that political expression may generate identification with or exposure to civil society actors, reinforcing trust in NGOs even as it appears to weaken trust in state institutions.
Figure 3. Coefficient estimates from ordinal regression model predicting Trust in NGOs. Each point represents the estimated effect of a variable, with horizontal bars indicating 95% confidence intervals.
Country specific findings
Across all models, education exhibits a consistent and negative association with institutional trust. Individuals with tertiary education (some college and beyond) are significantly less trusting of political, protective, and civic institutions, which may reflect heightened expectations or selective information exposure. Gender, income, and urban residence do not demonstrate consistent effects.
Country-level variation in trust is substantial across the models. Respondents in Vietnam, Singapore, and Indonesia generally report higher levels of political and protective trust. This is most apparent in the index models. Vietnam stands out with the highest levels of trust in both political trust (2.825, [2.064, 3.586]) and protective trust (0.872, [0.624, 1.120]). Singapore also displays significantly higher levels of trust, especially in political institutions (1.665, [1.028, 2.303]) and protective institutions (0.626, [0.438, 0.814]). Indonesia shows a similar, if more modest, pattern with significant positive coefficients in both trust indexes. These results suggest that in some authoritarian and hybrid regimes, trust in government institutions remains high, at least in relative terms.
By contrast, respondents in the Philippines and Myanmar report consistently lower levels of trust. In the political trust index, the Philippines shows a significant negative association (−0.664, [−1.315, −0.012]), and in Myanmar trust in protective institutions is sharply lower (−0.611, [−0.845, −0.376]). These patterns are mirrored in the NGO trust model. Respondents in Thailand (−1.261), the Philippines (−1.024), and Singapore (−1.015) show the largest negative coefficients. Malaysia (−0.913), Myanmar (−0.638), and Vietnam (−0.482) follow, all with confidence intervals that exclude zero. These findings underscore how trust in different institutions does not necessarily move in parallel. Vietnam and Singapore, for example, show high levels of trust in government and protective institutions but significantly lower levels of trust in NGOs. Conversely, countries like Malaysia and Thailand, where political trust is relatively average or inconsistent, report particularly low trust in civil society. Myanmar is an outlier in both respects, with low trust in NGOs and in protective institutions, likely reflecting the effects of political turmoil and repression following the military coup. The Philippines also shows low levels of trust across all three domains, pointing to deeper dissatisfaction or disillusionment with both government and non-government institutions.
This pattern aligns with earlier research suggesting that democratic institutions do not automatically produce higher levels of trust. Inoguchi (Reference Inoguchi2017) observes that across Asia, political trust tends to be lower in established democracies like Japan, South Korea, and Taiwan, whereas higher levels of trust are often found in more centralized or authoritarian regimes. Jiang and Zhang (Reference Jiang and Zhang2021) describe the phenomenon of hierarchical trust, in which citizens express high levels of trust across all levels of governance under authoritarian rule. They suggest that state control over political information and the use of propaganda may help sustain trust in political institutions. In such contexts, individuals who consume political news, whether through traditional or online channels, are more likely to express trust in government. These results reinforce the idea that institutional trust in Southeast Asia cannot be easily mapped onto regime type. Instead, institutional trust reflects both country-specific political dynamics and the structure of information control. Even when accounting for expressive political behavior and social media use, patterns of trust remain shaped by broader political context.
Wave specific findings
Additionally, across all models, respondents in Wave 5 expressed significantly higher levels of institutional trust compared to Wave 4. In the NGO trust model, the coefficient for Wave 5 is 1.235 (0.847, 1.622), indicating a marked increase in the odds of reporting greater trust in NGOs. Similar patterns appear in the index models: political trust increased (2.50, [1.19, 3.81]), and so did protective trust (0.95, [0.56, 1.34]). These are among the most substantial effects observed across all variables, suggesting a meaningful shift in attitudes between the two survey waves. The period captured by Wave 5 coincides with major regional developments that likely contributed to this change. Most notably, the COVID-19 pandemic began during the latter half of the fieldwork period. Across Southeast Asia, states adopted highly visible, state-centered interventions, such as curfews, border closures, contact tracing, and stimulus programs, which although sometimes unpopular, may have bolstered perceptions of state competence. Such responses may help explain the rise in protective trust.
At the same time, national-level events likely shaped trust in distinct ways. In Myanmar, the military coup in early 2021 and the violent suppression of protest movements likely contributed to the significant drop in trust in NGOs and protective institutions observed in the data. In Thailand, youth-led protests beginning in 2020 challenged military authority and were met with repression and emergency decrees. These conditions may have depressed trust in civil society and democratic institutions, particularly among younger respondents. Malaysia faced a prolonged period of elite-level political instability between 2020 and 2022, marked by coalition breakdowns and a state of emergency, which may have undermined confidence in political institutions. In Indonesia, protests over anti-corruption laws and broader democratic setbacks reflected tensions between an increasingly mobilized public and state institutions, though trust in political institutions remained relatively stable. Taken together, these dynamics underscore that the increase in trust during Wave 5 cannot be interpreted simply as a product of economic recovery or routine institutional performance. Instead, the observed changes appear to reflect country-specific responses to crisis, regime consolidation, or elite contestation. Whether these shifts in trust are durable remains uncertain.
Robustness checks
Is there a general negative association between social media use and trust? It is possible that social media lowers trust broadly, including interpersonal trust. To examine this, I analyzed two ABS questions measuring trust in one’s neighbors (Q25) and in other people (Q26). I also considered whether social media reduces trust in mainstream media institutions, such as television (Q53) and newspapers (Q54), which might be conflated with traditional governance institutions. The results show no consistent negative relationship between social media use and trust in these four domains. By contrast, individuals who express their political opinions on social media report significantly higher trust in other people and in television news. These findings imply that political engagement on social media may selectively reinforce certain forms of institutional and interpersonal trust rather than cause a general decline. Full results are reported in the Appendix (Table 8).
Is this an effect that is particular to traditional governance institutions? I conduct ordinal regressions for variables on trust in election commissions and civil servants. While these bureaucratic bodies are connected to the government, they are often separate entities but may still be conflated with political institutions. The models indicate that political expression is associated with a reduction in trust in these institutions (available in the Appendix: Table 5). Therefore, the effect remains consistent across traditional governmental institutions.
Given Southeast Asia’s political diversity and varying social media access, pooling data across countries can introduce heterogeneity that affects results. To address this, the main analysis uses fixed effects models with country dummies, isolating within-country variation and controlling for unobserved country-level factors. Additionally, to assess potential variation across survey waves, which may capture shifting political climates or platform usage patterns not fully controlled for, models were also estimated separately for each wave using their original scales. These produce largely consistent patterns of association, though the wave 5 model for the protective institutions index shows no significant relationship with social media usage and political expression (Appendix, Tables 3 and 4). This exception suggests some temporal variation, but overall, the core relationships are not driven solely by wave-level effects.
In addition to this, to ensure the robustness of these findings, additional analyses were conducted. Generalized linear mixed models controlling for GDP and democracy were also estimated, producing substantively similar results (Appendix: Tables 5 and 6). Jackknife tests excluding one country at a time confirm that no single country disproportionately influences the results (Appendix: Table 7). Together, these checks provide confidence that the observed associations reflect genuine within-country relationships and are not driven by cross-country or wave-level heterogeneity.
However, while the findings demonstrate robust associations between social media usage and trust in institutions, there is an acknowledgement of the inherent limitations of cross-sectional data in establishing causality. The observed relationships may reflect selection effects, wherein individuals with pre-existing dispositions, such as higher trust in civil society organizations or skepticism toward governance institutions, may self-select into social media engagement. Although the models control for socioeconomic factors, unobserved confounders could influence both social media adoption and trust outcomes. To mitigate these concerns, future research should employ methodologies that enhance causal identification. Leveraging exogenous variation in social media access could isolate causal effects. Exogenous shocks, such as sudden platform bans (e.g., Facebook’s temporary suspension in Vietnam in 2016) or regulatory changes, could provide quasi-experimental settings. These approaches, while beyond the scope of the current data, would strengthen causal claims and clarify mechanisms linking social media to institutional trust. The findings nevertheless provide critical baseline evidence for future investigations in hybrid and authoritarian contexts.
Discussion
The findings indicate that political expression on social media is associated with significant changes in trust in governmental institutions and NGOs. Two mechanisms help explain this pattern. First, social media offers access to information that is not available through traditional media channels. Individuals can encounter critiques of state narratives, alternative news sources, and grassroots commentary that would otherwise remain inaccessible, particularly in environments with tight media control. This helps account for the regional trend of declining institutional trust. Second, social media allows civil society organizations to engage with the public more effectively. NGOs use these platforms not only to disseminate content but also to build interactive relationships that convey transparency and responsiveness.
Therefore, social media functions as a venue for civic engagement, enabling individuals to interact with advocacy groups, access alternative information, and participate in online discussions. These findings are consistent with prior research showing that these platforms give marginalized voices tools for expression and enable grassroots mobilization even where civic space is restricted (Weiss and Aspinall, Reference Weiss and Aspinall2012; Lim, Reference Lim2023). Citizens use these platforms to voice dissatisfaction and challenge official narratives, which correlates with declining institutional trust across Southeast Asia. Previous studies have shown that social media platforms amplify exposure to government failures, including corruption, mismanagement, and citizen grievances, contributing to political cynicism (Sinpeng and Tapsell, Reference Sinpeng and Tapsell2020). This effect may be intensified in Southeast Asia due to varying degrees of media control, prompting reliance on social media for unfiltered political content.
The findings also align with studies showing how NGOs in the region leverage social media to humanize their missions and build credibility. Indonesian environmental groups have used Instagram to document reforestation efforts, increasing public trust by providing visual evidence of their work (Wahid, Reference Wahid2024). Malaysian NGOs engage with TikTok to challenge state misinformation on indigenous land rights, positioning themselves as credible information sources (Sinpeng and Tapsell, Reference Sinpeng and Tapsell2020). These cases reflect a shift in how NGOs cultivate legitimacy through content that demonstrates accountability and engages users directly. While governments attempt to control narratives, NGOs are using the same platforms to project transparency and responsiveness.
These results must also be understood in the context of a broader global decline in institutional trust. Public dissatisfaction is increasing across regime types, and support for democracy itself is weakening (Brady and Kent, Reference Brady and Kent2022; Citrin and Stoker, Reference Citrin and Stoker2018; Diamond, Reference Diamond2022). Social media does not necessarily cause this erosion but provides a space where grievances, disillusionment, and anti-establishment sentiment circulate. While much existing research focuses on right-wing populism in Western settings, Southeast Asia faces its own variant of digital extremism, especially in religious contexts. In Malaysia, TikTok has become a key site for the spread of extremist content, and many recent takedowns relate to violations of laws on religious and ethnic tolerance (Saidin and Azrun, Reference Saidin and Azrun2024).
The broader political transformation in the region reflects a decline in traditional authority and a turn toward alternative sources of legitimacy. States have responded with greater regulation and censorship, but these measures often deepen public distrust rather than restore it (Lim, Reference Lim2023). The corresponding rise in trust in NGOs illustrates this shift. Disillusionment with state institutions has led many individuals to seek representation, accountability, and responsiveness through civil society organizations. This dynamic can be interpreted through Hirschman’s (Reference Hirschman1970) “exit, voice, and loyalty” framework, in which disaffected individuals either disengage, express dissatisfaction, or remain committed despite grievances. In this setting, NGOs provide a form of “exit” from formal political channels, acting as alternative venues for expression and collective action (Lindenberg, Reference Lindenberg1999; Zuhlke, Reference Zuhlke2022).
However, while some versions of government failure theory suggest that NGOs grow in prominence when states are weak or unresponsive (Weisbrod, Reference Weisbrod1977; Salamon, Reference Salamon1987), empirical work complicates this view. Many NGOs depend on state support and legal frameworks for their operations. Lecy and Van Slyke (Reference Lecy and Van Slyke2013) show that government funding significantly contributes to nonprofit expansion, and Paarlberg and Zuhlke (Reference Paarlberg and Zuhlke2019) find that government capacity often correlates with the size of the nonprofit sector. These findings suggest that trust in NGOs does not necessarily reflect a rejection of the state. Many NGOs collaborate with governments, even in authoritarian settings, negotiating limited space for advocacy and service delivery (Hansson and Weiss, Reference Hansson and Weiss2024). Teets (Reference Teets2014) describes how consultative authoritarianism allows NGOs to shape policy informally. These layered arrangements imply that trust in NGOs reflects a redefinition of political legitimacy rather than a simple withdrawal from state institutions. What emerges is not a collapse of trust in the state, but a diversification of political authority in an era of digital connectivity and uncertainty.
Conclusion
This study has examined the relationship between social media and institutional trust, showing that politically active social media users tend to express lower levels of trust in governmental institutions and higher levels of trust in NGOs. Social media enables access to alternative sources of information and provides a venue for community building and civic expression, even in restrictive regimes. However, despite a resilient civil society in parts of Southeast Asia, research on the nonprofit sector remains limited. Given rising educational attainment, a growing middle class, and high rates of internet use among youth, the NGO sector is poised to expand its reach and relevance. This study contributes to growing literature on civil society in authoritarian and hybrid regimes and to scholarship on the emerging “third sector” in Asia.
This analysis, however, has several limitations. The use of cross-sectional survey data from the Asian Barometer restricts causal inference and captures snapshots. The cross-national scope introduces variation in political systems, digital environments, and civic traditions that cannot be fully accounted for. Country-level studies could generate more theoretically precise insights into how regime type, state capacity, and patterns of internet access affect institutional trust. Additional research using the Asian Barometer might explore whether other factors, such as attitudes toward globalism or traditionalism, shape trust in governance or civil society. Identifying the drivers of NGO trust may help nonprofits and their supporters better understand how to engage donors and volunteers across the region.
Furthermore, a significant data limitation concerns the lack of detail on specific individual social media behaviors. Knowing how users engage with content and organizations on these platforms is essential, as social media can enable both civic resistance and authoritarian control. Grassroots groups, such as the Prey Lang forest community in Cambodia, have used digital tools effectively to mobilize public support and achieve political outcomes. Their 2012 online campaign, supported by urban youth, led to the designation of their ancestral lands as a national wildlife sanctuary (Young, Reference Young2021). At the same time, digital spaces can be manipulated by uncivil actors. Mass disinformation campaigns and platform surveillance are not addressed in this paper, but they remain central to the broader dynamics of trust. Digital and media literacy can determine whether social media promotes civic engagement or exacerbates fragmentation. Evidence from Myanmar shows that low digital literacy within a mobilized online public can contribute to intercommunal conflict and ethnic violence (Rio, Reference Rio2021). These findings suggest that while social media provides an important space for civic interaction, it also carries risks when exploited by authoritarian actors.
The long-term effects of social media on institutional trust remain uncertain. Longitudinal research is needed to understand how trust evolves with sustained exposure to social media. Future work should also investigate the content and interactions that shape user attitudes toward institutions. Better understanding of these mechanisms can inform both policy and practice. For NGOs and their donors, this means developing communication strategies that build trust through transparency, responsiveness, and consistent engagement. Supporting digital literacy initiatives may also be key to strengthening civil society while limiting the corrosive effects of misinformation and state manipulation.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0143814X25100834.
Data and code availability
Replication materials are available in the Journal of Public Policy Dataverse at https://doi.org/10.7910/DVN/7FPDAJ. Parr, Christianna Sirindah, 2024, “Does Social Media Undermine Trust? Institutional Trust in Civil Society and Governance Institutions.”
Acknowledgements
The author thanks the anonymous reviewers for their excellent feedback. Many thanks to Aseem Prakash for his invaluable comments and insight.
Funding statement
The author received no financial support for the research, authorship, and/or publication of this article.
Competing interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Open access
