1. Introduction
The proliferation of authoritarian ideologies in regions afflicted by terrorism threatens democratic resilience and the consolidation of peace. The nexus between terrorism, authoritarianism, and democratic stability constitutes a critical challenge in contemporary conflict studies. As authoritarian ideologies gain traction in politically unstable areas plagued by terrorism, core democratic values—such as individual rights, political participation, and civil liberties—face escalating peril. Autocratic leaders often exploit the vulnerabilities and fear generated by terrorism to garner public support and entrench their power (Applebaum, Reference Applebaum2024). Moreover, individuals exposed to the trauma of terrorism are frequently more inclined to endorse non-democratic regimes, perceiving them as more effective in addressing security threats (Merolla and Zechmeister, Reference Merolla and Zechmeister2019).
Promoting democratization has the potential to combat terrorism, as democracy addresses the root causes of conflict—such as ethnic discrimination—and provides pathways for lasting conflict resolution (Windsor, Reference Windsor2003). Empirical evidence supports the comparative advantage of democracies in countering terrorism and fostering peace (Magen, Reference Magen2018).Footnote 1 However, despite these findings, the belief in an autocratic advantage in combating terrorism persists, compounded by democratic backsliding in countries facing persistent threats (Hetherington and Suhay, Reference Hetherington and Suhay2011). This global phenomenon has raised significant concerns among democracy advocates and international donors, prompting critical questions about how to reverse democratic decline while simultaneously promoting peace and political stability.
Among the various strategies for promoting democracy, resolving conflicts, and reducing the appeal of extremist ideologies (Ghosh et al., Reference Ghosh, Chan, Manuel and Dilimulati2017), democracy promotion interventions targeting ordinary citizens stand out as a promising solution.Footnote 2 To advance democratic resilience and consolidation, international donor organizations have allocated significant resources to supporting democracy through civil society initiatives and civic education programs (Finkel, Reference Finkel2002, Reference Finkel2014; Hyde et al., Reference Hyde, Lamb and Samet2023). Beyond traditional, in-person initiatives, research has shown that democratic promotion using online educational programs can also effectively reinforce preferences for democratic governance (Neundorf et al., Reference Neundorf, öztürk, Finkel and Ramírez2025). Such interventions have been found to reduce approval of authoritarian governance, increase the likelihood of voting for pro-democratic candidates, enhance knowledge of liberal democratic principles, and encourage greater political engagement (Ferrali et al., Reference Ferrali, Grossman and Larreguy2023; Finkel et al., Reference Finkel, Neundorf and Rascón Ramírez2024).
Despite the substantial body of research on the efficacy of both in-person and online educational programs in promoting democratic values and behavior, a gap persists in understanding its impact in regions afflicted by chronic terrorism, where authoritarianism is on the rise. The dual challenge of fostering peace while safeguarding democracy is particularly acute in conflict-affected environments. While promoting democracy through educational programs proved effective in diminishing the appeal of violent extremism (Ghosh et al., Reference Ghosh, Chan, Manuel and Dilimulati2017; Krause et al., Reference Krause, Gustafson, Theriault and Young2022) and bolstering democratic support, its effectiveness in environments of acute terrorism and autocracy—where immediate threats often justify authoritarian responses—remains uncertain. Specifically, critical questions remain unanswered: (1) Can online democratic promotion interventions, which reach diverse and remote audiences, effectively counter the detrimental effects of terrorism? (2) What types of intervention content are most impactful in these settings? (3) How do individuals’ experiences with terrorism—whether through direct exposure, residence in high-risk areas, or personal loss—influence the effectiveness of these interventions?
This paper investigates these questions by assessing the potential of online democracy promotion programs in Burkina Faso, a West African country grappling with chronic terrorism and authoritarian sentiment. Using a well-powered randomized online experiment, participants were exposed to one of three treatments: (1) a general civic rights video, (2) a video emphasizing democracy’s role in countering terrorism, (3) a video emphasizing democracy’s role in countering terrorism within the specific context of Burkina Faso, or (4) a placebo group. The study measured the interventions’ effects on key democratic attitudes and behaviors.
The findings reveal that the treatments increased support for democracy and the rejection of authoritarian alternatives in terrorism-affected contexts. The treatment emphasizing the universal advantages of democracy in addressing terrorism had the strongest impact, while the context-specific treatment proved ineffective. This suggests that emphasizing universal democratic advantages may resonate more in chronic terrorism environments than tailoring content to local experiences. The interventions also positively shaped perceptions of democracy as a secure and effective system for combating terrorism and restoring peace. However, these effects are not contingent upon the respondents’ lived experience of terrorism, whether direct or indirect.
This research contributes to the scholarship on conflict resolution and democratic resilience by investigating how targeted democratic promotion initiatives can sustain democratic support in fragile, conflict-affected states. By addressing the specific challenges posed by terrorism, the study highlights the role of education in fostering democracy and countering authoritarianism among ordinary citizens.
2. Terrorism, authoritarianism, and democratic resilience
Citizen support is the cornerstone of democracy, sustaining its legitimacy and guiding its consolidation (Claassen, Reference Claassen2020). This is particularly vital in fragile states like Burkina Faso, where terrorism threatens both collective bonds and individual safety. In such environments, nurturing public belief in the value of democracy becomes crucial, as citizens who view it as the best form of governance are more likely to defend it (Bratton and Mattes, Reference Bratton and Mattes2001).
Terrorism, a deliberate tactic designed to spread fear and destabilize societies (Baldwin et al., Reference Baldwin, Ramaprasad and Samsa2008), poses a significant challenge to democratic governance. It often prompts individuals to embrace authoritarian attitudes, favoring measures like media censorship, centralized power, and autocratic leadership (Merolla and Zechmeister, Reference Merolla and Zechmeister2009; Hetherington and Suhay, Reference Hetherington and Suhay2011). When people feel threatened, they tend to gravitate toward “strongman” leaders, who are perceived as capable of offering swift and decisive solutions (Laustsen and Petersen, Reference Laustsen and Petersen2017; Laustsen, Reference Laustsen2021). Authoritarian regimes frequently exploit this fear, using terrorism as a justification to tighten their grip on power and curtail civil liberties (Applebaum, Reference Applebaum2024).
Although some argue that authoritarian governments manage better terrorism, the evidence paints a more nuanced picture (Wade and Reiter, Reference Wade and Reiter2007; Chenoweth, Reference Chenoweth2013; Gaibulloev et al., Reference Gaibulloev, Piazza and Sandler2017). Democracies outperform authoritarian regimes by employing inclusive strategies that address security threats while respecting individual rights (Magen, Reference Magen2018). These approaches foster long-term peace, stability, and social cohesion, in contrast to the repressive methods typical of autocracies. This paradox—where public preference for authoritarianism persists despite its practical shortcomings—highlights the urgency of countering autocratic narratives, particularly in fragile states with a high probability of terrorist attacks.
Terrorism activates deep-seated psychological responses, as outlined in terror management theory, by amplifying people’s desire for safety, order, and predictability (Pyszczynski et al., Reference Pyszczynski, Greenberg and Solomon1999; Pyszczynski, Reference Pyszczynski2004; Phillips, Reference Phillips2010). This presents a crucial opportunity to reframe democratic governance not only as capable of safeguarding lives but also of protecting civil liberties. This opportunity arises because chronic terrorist attacks expose governments’ inability to protect citizens, undermining public confidence and demoralizing populations (Spilerman and Stecklov, Reference Spilerman and Stecklov2009). This erosion of autocratic legitimacy creates an opening to position democracy as the most effective regime for conflict resolution and peace-building.
We argue that building public support for democracy in the face of terrorism requires educating individuals to critically assess the claims of authoritarian regimes while emphasizing democracy’s capacity to ensure both security and freedom. By directly addressing the failures of autocratic responses and presenting democracy as a viable alternative, we postulate:
Hypothesis 1 (H1): Democracy promotion interventions foster support for democratic systems and their ability to protect civil liberties and lives.
3. Framing democracy promotion
Promoting democracy in contexts marked by persistent terrorist attacks presents unique challenges, particularly in positioning democracy as a viable response to the pressing needs of the population. According to Neundorf et al. (Reference Neundorf, öztürk, Finkel and Ramírez2025), the impact of democratic promotion is most pronounced during the democratic developmental stages, where the demand for civic education is highest, and the potential for political change is most realistic. The challenges of democratic promotion are further compounded in autocracies, where foreign actors often lack sufficient leverage or political will, and domestic conditions may not be conducive to change (Hyde et al., Reference Hyde, Lamb and Samet2023; Donno, Reference Donno and Wolf2024).
But which arguments promoting democracy are best for maximizing its impact on citizens’ democratic attitudes? To explore which framing resonates most strongly with individuals living in chronic terrorism contexts, we developed three democratic promotion interventions, each targeting different levels of abstraction and psychological engagement. The first intervention is an abstract video highlighting core democratic values such as civic rights (1) and liberties while also emphasizing the risks of non-democratic systems. This video aims to promote general democratic ideals and contrast them with the potential dangers of authoritarianism without any specific reference to terrorism. The second intervention builds upon the first by focusing more directly on the “needs” and motivational congruence of respondents (Higgins, Reference Higgins1998), considering their lived experience of chronic terrorist attacks. However, this video presents a general argument about democracy’s effectiveness against terrorism, using brief examples from unrelated countries at the end to support the idea of democracy as a universal solution to terrorism in general (2). The third intervention is a context-specific video that directly engages with the realities respondents face in the context of chronic terrorism. It includes information derived from their daily exposure to terrorism in Burkina Faso (3) and examines the current regime’s counterterrorism measures. This framing aims to resonate more strongly with respondents by addressing their specific concerns and the pressing need for stability, peace and security, and a return to normalcy (Vázquez et al., Reference Vázquez, Pérez-Sales and Matt2006; Mannetti et al., Reference Mannetti, Brizi, Giacomantonio and Higgins2013). These three videos vary in abstraction and emotional distance, ranging from the more general and abstract (first video) to the more emotionally engaging and contextually grounded (third video).
4. Message framing and instrumental logic of democracy
Citizens in fragile and conflict-affected settings are often less responsive to appeals grounded in abstract democratic ideals, showing greater concern for practical solutions to the insecurity they face (Higgins, Reference Higgins1998; Mannetti et al., Reference Mannetti, Brizi, Giacomantonio and Higgins2013). As Higgins (Reference Higgins1998) explains, individuals are more likely to be persuaded when messages are aligned with their underlying motivational orientation. In high-threat contexts, this often means prioritizing safety, stability, and immediate relief over aspirational or normative content. Framing democratic systems as effective mechanisms for countering terrorism and restoring order may therefore challenge the perceived “autocratic advantage” in managing crisis (Merolla and Zechmeister, Reference Merolla and Zechmeister2019). This instrumental framing resonates with citizens’ need for protection and peace, particularly in societies where terrorism constitutes a dominant and persistent public concern. Accordingly:
Hypothesis 2a (H2a): In a context of chronic terrorism, a democratic promotion intervention that frames democracy as a solution to terrorism will have a stronger effect than a generic rights-based intervention.
5. Democracy promotion intervention and local relevance
Beyond instrumental framing, the perceived relevance and motivational tone of a message influence its persuasiveness. Regulatory fit theory posits that people are more likely to respond to messages that align with their current motivational state: Promotion-focused people seek ideals and gains, while prevention-focused people prioritize safety, and avoidance of losses (Higgins, Reference Higgins1998; Mannetti et al., Reference Mannetti, Brizi, Giacomantonio and Higgins2013). In terrorism-affected contexts, where prevention-oriented mindsets are likely to dominate, messages acknowledging local threats and offering context-specific solutions should be more effective. This suggests that democratic interventions explicitly referencing Burkina Faso’s own experience with terrorism, rather than speaking in generalities, may better resonate with citizens’ immediate concerns.
Hypothesis 2b (H2b): A context-specific democracy promotion intervention that addresses terrorism in Burkina Faso will have a stronger effect than a general message about the effectiveness of democracy.
6. Chronic threat and the limits of localized framing
However, the advantages of context-specific messages can be offset by the emotional and psychological consequences of chronic terrorism. As Spilerman and Stecklov (Reference Spilerman and Stecklov2009) show, prolonged exposure to violence does not simply intensify political receptiveness. In contrast, it can produce dissociative coping mechanisms, emotional disengagement, denial of the reality, or projection of blame on other people or distrust of others, which help people adapt to abnormal conditions but simultaneously reduce their openness to political communication. In such environments, even well-intentioned democratic appeals may be perceived as intrusive, emotionally taxing, or manipulative, especially if they resemble official state rhetoric or evoke familiar traumas.
This insight challenges the assumption that more tailored messages are necessarily more persuasive. Instead, general or abstract framings may achieve greater regulatory fit by offering symbolic distance and emotional protection. Such framings allow citizens to engage with democratic ideals without reliving the daily reality of threat, distrust, or disappointment.
Hypothesis 2c (H2c): In chronic terrorism contexts, general democracy promotion messages, focused on abstract democratic advantage in terrorism fight can be more effective than context-specific interventions.
7. Experiencing terrorism and the impact of democracy promotion
The literature on affective responses to violence from terrorism shows that affected individuals tend to express stronger emotional reactions, such as anger or anxiety, which can increase their receptiveness to political messages (Vasilopoulos et al., Reference Vasilopoulos, Marcus and Foucault2018; Marcus et al., Reference Marcus, Valentino, Vasilopoulos and Foucault2019). The frame of democracy as a credible and effective response to insecurity may therefore have stronger effects on this subgroup. However, the psychological toll of repeated trauma could also dull emotional sensitivity and reduce the effectiveness of even highly salient messages (Spilerman and Stecklov, Reference Spilerman and Stecklov2009).
This tension reflects a theoretical uncertainty: while conventional models expect heightened responsiveness among victims, the realities of chronic threat may instead lead to political fatigue and emotional numbness. We retain the conventional expectation that direct exposure increases message effectiveness, while remaining attentive to the potential for attenuation in chronic settings.
Hypothesis 3 (H3): Democracy promotion interventions will be most effective among individuals personally affected by terrorism.
8. Research design
To test our hypotheses, we implement a randomized controlled trial in Burkina Faso, testing the impact of online democracy promotion programs on key democratic attitudes and behavior. In this section, we outline our case selection and research design.
8.1. Case study Burkina Faso
Burkina Faso offers a critical case for examining democracy promotion amid chronic terrorism and autocratization. Its trajectory raises the challenge of designing interventions that respond to public demands for security (Higgins, Reference Higgins1998; Mannetti et al., Reference Mannetti, Brizi, Giacomantonio and Higgins2013) while accounting for psychological coping mechanisms that may blunt the impact of localized messages (Vázquez et al., Reference Vázquez, Pérez-Sales and Matt2006; Spilerman and Stecklov, Reference Spilerman and Stecklov2009).
Following independence from France in 1960, the country was governed largely by authoritarian regimes until democratic elections in 2015. However, escalating terrorist violence has since fueled political instability, resulting in two military coups and sharp democratic decline. The democratic breakthrough of 2014–2015, when a popular uprising ended Blaise Compaoré’s 27-year rule and opened the way for free elections, marked a brief peak of democratic enthusiasm. The 2014 revolution was the culmination of a long trajectory of prodemocracy mobilization. As Figure 1 illustrates, the liberal democracy index rose modestly during the 1990s but has collapsed since 2016, reaching closed autocracy by 2023 (0.18 on the V-Dem scale) (Coppedge et al., Reference Coppedge, Gerring, Knutsen, Lindberg, Teorell, Altman, Bernhard, Cornell, Fish, Gastaldi, Gjerløw, Glynn, Grahn, Hicken, Kinzelbach, Marquardt, McMann, Mechkova, Neundorf, Paxton, Pemstein, Rydén, von Römer, Seim, Sigman, Skaaning, Staton, Sundström, Tzelgov, Uberti, Wang, Wig and Ziblatt2024). This trajectory reflects both the initial optimism of the post-revolution transition and its erosion under growing insecurity and autocratic governance.
Liberal democratic index, 1960–2023 (V-Dem).
Successive governments used security narratives to justify curtailing civil liberties, restricting media outlets, such as the BBC and Voice of America, and detaining activists and political leaders (see BBC and Le Monde).
Before 2015, Burkina Faso remained largely insulated from the wave of terrorism affecting the wider Sahel, with Mali serving as the primary hotspot. Since then, the country has experienced a sharp escalation in violence, with its Global Terrorism Index rising from 3 in 2015 to 8.56 in 2022 (Figure A2, Online Appendix). This index ranges from zero to ten, with higher scores indicating greater impact (see Appendix A.2). The surge reflects mounting attacks by groups affiliated with Al-Qaeda and ISIS. In 2023 alone, more than 1,000 terrorism-related fatalities were recorded, part of a broader regional trend (Institute for Economics & Peace, Reference Institute for Economics & Peace2024). Figure 2 shows the spatial concentration of attacks between 2018 and 2022, particularly along the borders with Mali and Niger.
Spatial distribution of terrorism attacks, 2018–2022 (ACLED).
Terrorism and the rise of authoritarian narratives have eroded public support for democracy in Burkina Faso. Figure 3 shows the decline in support for democracy, as measured by the Afrobarometer survey, from 80% in 2014/15 to 55% in 2021/23 amid political turmoil and terrorism. At the same time, support for non-democratic alternatives surged from 4.4% to 25%. This shift, occurring around 2015–2016, suggests a negative impact of terrorism and autocratic security narratives on democracy.
Individual support for democracy, 2008–2022 (Afrobarometer Data).
Afrobarometer survey rounds: R4: September 2008, R5: October 2012, R6: April 2015, R7: October 2018, R8: October 2021, R9: September 2022.
8.2. Data collection
After we received ethical approval and registered our research design, we collected data between 2 November and 3 December 2024.Footnote 3 We recruited our research participants using paid advertisements on the Meta platforms (i.e., Facebook and Instagram).Footnote 4 Our advertisements have been shown to over 700,000 social media users in Burkina Faso, inviting them to participate in our survey for the chance to win one $200 voucher for a local supermarket. Those who clicked on our advertisement were directed to the consent page of the survey, which was programmed on Qualtrics.
Recruitment of research participants using social media advertisements has become increasingly common among comparative political scientists, as this method allows unmediated access to billions of people across the World (Neundorf and Öztürk, Reference Neundorf and öztürk2023). Especially in the Global South, social media advertisements can generate a more diverse pool than online panels of commercial companies, such as YouGov or Qualtrics, which are mainly formed of “professional survey takers” (Boas et al., Reference Boas, Christenson and Glick2020; Rosenzweig et al., Reference Rosenzweig, Bergquist, Pham, Rampazzo and Mildenberger2025).
To ameliorate the self-selection bias and reach a more diverse sample, we took two measures. First, we used the targeting tools on the Facebook Ads Manager to show our advertisements to people from different demographic groups along certain quotas for each of these groups (Zhang et al., Reference Zhang, Mildenberger, Howe, Marlon, Rosenthal and Leiserowitz2020).Footnote 5 Second, we used material incentives to encourage the participation of social media users with lower political interest and education levels and used minimal references to the political nature of our survey in the advertisement (Neundorf and öztürk, Reference Neundorf and öztürk2025).Footnote 6
Overall, we recruited 2,004 respondents who provided valid responses on at least one of our outcome variables.Footnote 7 To assess the representativeness of our sample, we compared it to weighted data from Afrobarometer Round 9 and the national census. Table A3, presented in the Online Appendix, shows that our sample is nationally diverse, approaching population benchmarks in terms of demographic, ethnic and religious representation of respondents. However, it especially over-represents individuals with higher levels of education. This coverage bias is a limitation that is common to most online surveys conducted in the Global South (Rosenzweig et al., Reference Rosenzweig, Bergquist, Pham, Rampazzo and Mildenberger2025). We replicate our analyses using post-stratification weights to alleviate the effects of this bias, and discuss this issue in more detail as one of the limitations of our study in the Conclusion.
8.3. Experimental design and treatments
We opted to trial the effectiveness of short, online educational videos promoting democracy. The advantage of using the format of videos compared to in-person events in the context of heightened conflict is to minimize possible security risks. Further, watching videos online can maximize the reach of these interventions to several million compared to the limited reach of traditional in-person civic education events.
For this, we randomly assigned participants to one of three treatment groups or one placebo group using simple randomization.Footnote 8 Participants in the experimental condition were shown short, animated videos, which differ in content and framing: (1) A video that focuses on the general advantage of democracy over non-democratic systems, with an emphasis on civic rights, (2) a video emphasizing the universal advantage of democracy in addressing the general threat of terrorism, and (3) a Burkina Faso context-specific video, explaining the advantages of democracy in addressing terrorism in Burkina Faso. A control group received a placebo treatment focusing on space exploration.
These videos are 3–4 minutes long and links to the videos are included in Online Appendix, section B. These videos promoting democracy are carefully designed to educate individuals about their rights, responsibilities, and roles as citizens within society. They aim to promote knowledge, skills, and values necessary for active and informed participation in civic and democratic life. We meticulously developed scripts for each video and created short videos with background music that reflected the intended content.Footnote 9
In terms of specific content, the first treatment focusing on civic rights highlights the superiority of democracies in safeguarding citizens’ rights compared to non-democratic regimes. Democracies protect citizens’ political, individual, and economic rights, allowing free expression, equality, and fair treatment. In contrast, non-democracies suppress dissent and enable exploitation. To illustrate how democracies empower and safeguard their citizens, unlike authoritarian regimes, the video concludes with a brief example comparing Lithuania (a democracy) and Belarus (a dictatorship). This illustration serves to support the general argument. These countries were selected precisely because they are not related to the local context, reducing possible bias and enhancing the universal applicability of the message.
This treatment is closest to the standard approach of promoting democracy in general without emphasizing the particular challenge a country faces. We include this treatment as it allows us to test whether promoting democracy in this general and abstract way works even in severe circumstances of terrorism. It also serves as evidence that the change in democratic attitudes is driven by the democracy promotion framing, not by factors associated with the terrorism framing of the other two treatments.
The second intervention aims to counter the narrative on authoritarian advantages against terrorism, arguing that democracies, in general, are more effective in fighting terrorism than non-democratic regimes thanks to accountable institutions, respect for rights, and international cooperation. Democracies break cycles of violence by allowing peaceful dissent and preventing extremism through openness. Parallel to the country comparison presented in the first treatment video, the video concludes with a brief example comparing Lithuania (a democracy) and Belarus (a dictatorship) as examples to present evidence that democracies prioritize security while safeguarding freedoms, unlike authoritarian regimes.
Finally, the third treatment focuses on the current security and political situation in Burkina Faso and the prospect of a democratic system addressing the challenge of terrorism in the country. Here, we customized the content by incorporating localized facts, real video footage, a native voice-over,Footnote 10 and background music tailored to the local context. The goal was to present a video that resonates with the respondents’ realities and environment while following the same arguments as included in the second treatment.
In the video, we argue that non-democratic regimes like Burkina Faso provide fertile ground for extremism due to institutional instability and unchecked state violence, such as the 2023 massacre of civilians by state forces in the village of Karma.Footnote 11 Burkina Faso could address these challenges by embracing democracy, which prioritizes security and citizen welfare. In this treatment, we selected Ghana as the comparative case with Burkina Faso (and not Belarus versus Lithuania as in the other two treatments) due to its strong cultural ties and its status as a neighboring country with a well-established reputation for democratic governance.Footnote 12
All treatment videos followed the same structure, balancing positive and negative arguments and incorporating empirical evidence for comparability. They were designed to foster democratic support while avoiding being overly persuasive or insufficiently engaging. Research reveals that the public may develop resistance to advertisements that are perceived as too “pushy,” yet they are unlikely to engage with messages that are not sufficiently direct (Hewitt et al., Reference Hewitt, Broockman, Coppock, Tappin, Slezak, Coffman, Lubin and Hamidian2024).
A validity concern is that effects might reflect design features (voice, style) rather than content, with more educated respondents responding better to a neutral Western-style voice. Balance tests show no education differences across treatment arms (Figure E1), and conditional effects by education (Appendix N.3) reveal no systematic heterogeneity. This supports the view that effects stem from message framing rather than design features.
9. Outcome variables
We use six main outcome measures to assess the impact of democracy promotion interventions on democratic support and behavior. The outcomes are divided into three main groupsFootnote 13:
1. Perception of democracy’s efficacy in the context of terrorism: This outcome evaluates whether respondents believe democracy is effective in addressing terrorism, aiming to correct the misconception of an autocratic advantage in managing such threat.
2. General support for democracy and rejection of its alternatives:
• Support for democracy: This measure captures the abstract preference for democracy by asking respondents to self-report how important it is for them to live in a democracy.
• Rejection of authoritarian alternatives: This is a composite measure that evaluates respondents’ attitudes toward various non-democratic systems, including strongman rule, single-party rule, military rule, and religious law (e.g., Sharia law). Respondents were asked to assess how good or bad they perceived these systems to be.
3. Behavioral intention variables: These measures capture whether the democracy promotion interventions influence pro-democratic behavioral responses. Examples include intentions to donate to a pro-democracy NGO, participate in pro-democracy protests, or vote for a democratic candidate.
All outcomes were measured after respondents watched the videos. Table 1 presents the descriptive statistics, where the outcomes have been standardized to range from 0 to 1 to facilitate subsequent comparative analyses. A value of 0 indicates disagreement, while 1 signifies agreement. In our sample—across all treatments and placebo groups—74.8% prefer to live in a democracy, demonstrating the strong demand for democracy. However, we also see that only 46.6% reject authoritarian alternatives, which demonstrates the tension in the political narrative in contexts of terrorism.
Descriptive statistics: standardized outcomes
10. Results
We test our hypotheses using simple linear regression models. All models control for the pre-treatment covariates: gender, age, education, religion, ethnicity, preference to live in a democracy, voting intention for the current president, and political interest. In the first set of results, we combined the educational treatments—general civic rights treatment (T1), general terrorism treatment (T2), and Burkina Faso terrorism treatment (T3), and compared this group to the placebo group to test the overall impact of online democracy promotion on our main outcomes. This tests Hypothesis 1. The second set of analyses estimated the separate effects of each treatment group (T1, T2, and T3) compared to the placebo group. This tests Hypotheses 2a and 2b, focusing more specifically on the content of the different videos. The results are summarized in Tables 2 and 3. Lastly, we present results that condition the pooled treatment effect on the lived experience with terrorism of the respondents. The results of these analyses, testing Hypothesis 3, are presented in section 4.3.
Main treatment effects: All treatment groups versus placebo
Significance: *p < 0.05. Notes: The table presents OLS estimates and standard errors in brackets. The dependent variables were standardized to range from 0 to 1. (1): Democratic advantage in addressing terrorism, (2): Abstract democratic support; (3): Rejection of nondemocratic alternatives; (4): Vote pro-democratic candidate; (5): Donate to pro-democratic NGO; (6): Attend a pro-democratic protest. All models control for the pre-treatment variables: gender, age, education, religion, ethnicity, levels of democratic preferences, voting intention, and political interest. Table F1 in the appendix further reports the full results with all pre-treatment coefficients. Similar results are obtained when holding all observations constant across the outcomes (see Appendix, Table L1). We also reported the results using original scale and interval confidence (see Appendix, Table I1).
Comparing treatment groups versus placebo
Significance: *p < 0.05. Notes: The Table presents OLS estimates and standard errors in brackets. The dependent variables were standardized to range from 0 to 1. M1: Democratic advantage in addressing terrorism, M2: Abstract democratic support; M3: Rejection of nondemocratic alternatives; M4: Vote pro-democratic candidate; M5: Donate to pro-democratic NGO; M6: Attend a pro-democratic protest. All models control for the pre-treatment variables: gender, age, education, religion, ethnicity, levels of democratic preferences, voting intention, and political interest. P-values are calculated based on t-tests, where we compare the coefficients of civic rights with the general treatment and the Burkina Faso treatment. We also compare the general with the Burkina Faso treatment. Table F2 in the appendix further reports the results with all pre-treatment coefficients. We also reported the results using original scale and interval confidence (see Appendix, Table I2).
10.1. Main treatment effects
Table 2 presents the pooled treatment effects compared to the placebo. First, the interventions improved public perception of democracy’s ability to manage terrorism by 4.9 percentage points (M1), about 9% more than the placebo group. This outcome most directly relates to the content of two of the treatment videos related to terrorism and the results confirm that we could successfully enhance the perception of democracy’s advantage in dealing with terrorism compared to autocracies. We interpret this as the first necessary condition to further positively impact other democratic attitudes and behavior.Footnote 14
Second, we find that exposure to the treatments increases respondents’ stated importance of living in a democracy by 3.3 percentage points (M2), and a 3.7 percentage point increase in the rejection of authoritarian alternatives (M3). In both cases, this represents an increase of approximately 5% and 8% compared to the placebo group. Still, the treatments do not significantly influence pro-democratic behaviors.
Overall, promoting democracy through online educational interventions is more effective at enhancing abstract support for democracy and rejection of its alternatives than at fostering pro-democratic behaviors. We hence find mixed evidence in support of Hypothesis 1.Footnote 15
We further replicate our results by only controlling for unbalanced pre-treatment variables (gender, democratic preferences), using constant observations across all outcomes, and applying weighted data (see Online Appendix, sections G, L, H). In all cases, the results remain consistent with the main findings.
10.2. Comparing the content of democracy promotion interventions
Next, we examine the impact of our democracy-promotion interventions by comparing the effects of three online treatments (civic rights-T1, general terrorism-T2, and Burkina Faso terrorism-T3) to the placebo group. Table 3 shows the impact of each treatment compared to the placebo group. The civic rights treatment (T1) significantly affects two outcomes: public preference for democracy, increasing by 4.8 percentage points (M2), and rejection of non-democratic alternatives, increasing by 4.0 percentage points (M3). These effects, respectively, are about 7% and 9% higher than the control group.
The general terrorism treatment (T2) shows significant effects on most outcomes (5 out of 6). It notably boosts the perception of democracy’s ability to address terrorism threats, increasing this view by 8.7 percentage points (M1), or 15% more than the placebo group.
Additionally, the treatment raises preference for democracy by 3.5 percentage points (M2) and rejection of authoritarianism by 4.7 percentage points (M3), a 5% and 10% improvement compared to the placebo group. The biggest effect is seen in voting behavior, with a 5.2 percentage point increase in the likelihood of voting for a democratic candidate (M4), a 17% higher chance than untreated respondents. Moreover, the treatment boosts participation in pro-democratic protests by 4.7 percentage points (M6), 9% more than the control group.
Like the general terrorism treatment, the Burkina Faso treatment (T3) increased public perceptions of democracy’s ability to address terrorism by 3.7 percentage points (M1), a 7% higher impact than the placebo group. However, this effect is significantly weaker than the general terrorism treatment (t-test; p-values in Table 3). Contrary to expectations (H2b), respondents exposed to context-specific videos showed no significant changes in preferences, attitudes, or behavior compared to the placebo group.
We also provide an item-level analysis of authoritarian alternatives. Rights-based messages increase rejection of military and one-party rule, while the general frame works only against one-party rule. The Burkina Faso–specific frame shows no effect, consistent with evidence that over-localization can weaken democratic appeals. None of the interventions shift religion-based preferences (Sharia) (see Online Appendix, Table J1). These patterns align with Burkina Faso’s political history: nearly 27 years of one-party rule under Blaise Compaoré’s (1987–2014), followed by mass protests that ousted military regimes after coups in 2015 and 2022. Taken together, the results suggest that principled democratic messages resonate with the public’s demand for greater democratic participation.
Next, we compare the effect of the universal rights intervention (civic rights-T1) with the pooled terrorism interventions (T2/T3), hypothesizing that T2/T3 will have a stronger impact (H2a). The findings in Online Appendix Table G3 show that, in one of six outcomes, the intervention tailored to both terrorism issue and country-specific context is more effective than the universal rights treatment. Specifically, it has twice the impact in improving public perceptions of democracy’s ability to address terrorism threats, which confirms H2a.
Finally, we compared the two terrorism-focused treatments (Table 3). The joint probability test (T2 vs T3) shows significant differences in democratic advantage (M1) and participation in pro-democratic protests (M6). The democracy-promotion intervention emphasizing the universal benefits of democracy in combating terrorism had a significantly stronger impact on these outcomes than the country-specific intervention, testing H2c. Specifically, its effects were about twice as strong for the perception of democratic advantage and nine times higher for protest participation compared to the country-focused treatment (T3).
The fact that the context-specific treatment did not yield stronger effects requires further investigation. We asked respondents whether the videos were informative, representative of their country, and whether they would share them. Focusing on the three treatment groups (excluding the non-political placebo), Table 4 shows that the Burkina Faso-specific video (T3) received the lowest ratings across all three dimensions compared to the civic rights video.
Treatment effects: Terrorism treatment groups (T2/T3) versus civic rights (T1)
Significance: *p < 0.05. Notes: The Table presents OLS estimates and standard errors in brackets. The dependent variables were standardized to range from 0 to 1. M1: Video informative, M2: Video like country; M3: Share video with others; All models control for the pre-treatment variables: gender, age, education, religion, ethnicity, levels of democratic preferences, voting intention, and political interest. Table M2 presents the interaction between the pooled treatment and political interest. The results show no effect.
Similar results are found when comparing each treatment group to the placebo (see Online Appendix Table M1). In settings of chronic violence, people often cope by downplaying reality and distrusting others. This coping style, as Spilerman and Stecklov (Reference Spilerman and Stecklov2009) note, can blunt the impact of political messages, especially those seen as externally imposed or emotionally intrusive.
We conducted robustness checks to address potential empirical concerns, particularly whether the videos, which discuss terrorism, might prime respondents on security issues rather than democracy. If this were the case, it would be hard to separate the effect of the democratic message framing from respondents’ heightened security concerns, which could independently affect their democratic preferences, attitudes, and behaviors. To rule out this possibility, we asked respondents about their level of concern regarding Burkina Faso’s future security after watching the videos.Footnote 16 We then used their responses as an outcome variable. Our findings show that the videos do not increase respondents’ concerns about the country’s future security (see Model M4 in Online Appendix Tables K1 and K2).
This result suggests that our democracy promotion interventions are primarily effective in shaping respondents’ regime preferences in the context of terrorism rather than priming concerns about security issues. Given that the main arguments of each video center on the ability of democracies to ensure civic rights (whether in connection with terrorism (T2 + 3) or not (T1)), the results presented in Table 3 suggest that integrating terrorism-related content into (non-context specific) democracy promotion interventions makes democracy appear more relevant or effective in the face of terrorism.
This still leaves the question, of why the Burkina Faso-specific treatment (T3) had weaker effects than the general terrorism video (T2). For this, we turn to the qualitative analysis of open-ended feedback from the pilot. Respondents were asked to comment on the video they watched with the question: “Overall, what do you think of the video you just watched?” Most respondents had favorable perceptions of the videos, with mixed feedback for the Burkina Faso video (see Online Appendix, Section D, for the word cloud of all responses for each video). Individuals in the civic rights treatment group (T1) showed stronger support for democracy and rejection of non-democratic alternatives (Tables 2 and 3). Feedback suggests that emphasizing general democratic rights fosters critical thinking and reinforces democratic ideals, as seen in comments like: “It helps further develop everyone’s critical thinking,” and “This questionnaire is very instructive and helps to better understand the difference between a democratic and a non-democratic country.”
Second, respondents in the general terrorism group (T2) reacted positively to the treatment, with increased support for democracy framed as a solution to issues like terrorism. They valued the link between governance and security. Here are some examples of these sentiments: “It is good to study these situations, but one should not already take a position,” “Very instructive,” and “This questionnaire seems like an interesting opportunity to reflect on the concepts of democracy and the current political regime in Burkina Faso,” which indicates that people still related the content of the general terrorism video to their own context, despite Burkina Faso not being mentioned in the video.
In contrast, the Burkina Faso group results show mixed reactions of the pilot respondents, likely due to several factors. While some saw the relevance of democracy, others viewed the message as biased or impractical, as reflected in statements like: “I think this questionnaire tries to help us choose the form of governance we would like for our country,” “I feel this questionnaire questions the current situation in Burkina Faso, which is actually better in governance than the previous democratic regimes,” “I know this questionnaire aims to revolt against the population,” and “I think the questionnaire is biased.”
These responses emphasize the role of context and message framing in shaping democratic attitudes. Universal messages promoting core democratic principles (the civic rights—T1 and general terrorism—T2—treatments) generated more positive responses to the open-ended question. When political messages explicitly reference the national crisis, as in the T3 video, they may inadvertently trigger defensive reactions or skepticism, particularly if perceived as echoing foreign agendas.Footnote 17 This contrasts with the general terrorism video (T2), which, by offering symbolic distance and avoiding direct confrontation with traumatic local realities, may have created a psychologically safer space for reflection. The backlash against the T3 message, including accusations of bias and manipulation, highlights how chronic terrorism, the popularity of the regime,Footnote 18 and anti-Western sentiment intersect to shape receptiveness. In this context, abstract or universal framings, like those in the civic rights and general terrorism treatments, may achieve greater resonance precisely because they avoid activating deep-seated resistance or trauma (Spilerman and Stecklov, Reference Spilerman and Stecklov2009).
10.3. Heterogeneous treatment effects: lived experience and perceived safety
We investigate whether the pooled treatment effects vary by respondents’ experience with terrorism, measured using two indicators.Footnote 19 Results are based on interactions between the pooled treatment and these indicators. Marginal effects are plotted below; full estimators are in the Online Appendix, section N.1. The first measure captures direct experience with terrorism, based on whether they have encountered any of the following due to terrorism: (1) caring for a displaced person, (2) being injured or losing someone close, or (3) leaving home. Respondents with at least one experience were coded 1, others 0.Footnote 20 Figure 4 shows no significant conditional effects, as confidence intervals overlap.
Marginal effects of the democracy promotion treatment on democratic support by lived experience.
Disaggregated results confirm this pattern (Online Appendix, section N.2). The second measure, a socio-tropic indicator, assesses respondents’ proximity to terrorist attacks, providing an objective exposure measure. Using ACLEDFootnote 21 and Afrobarometer geocoded data on respondents’ residence,Footnote 22 we calculated each respondent’s distance to the nearest attack. This continuous measure was used as a moderator (H3), expecting stronger effects for those closer to attacks. Figure 5 shows no significant conditional effects, as indicated by overlapping confidence intervals. Non-linear models yield similar results (Online Appendix, Table N2).
Spatial exposure: Proximity to terrorist attacks.
In sum, neither proximity to violence nor personal exposure, nor related emotions (see the Online Appendix, Section O), moderated treatment effects, leading us to reject Hypothesis 3.
The near-universal salience of terrorism in Burkina Faso (77% reported direct exposure) may leave little room for individual variation. Moreover, as Spilerman and Stecklov (Reference Spilerman and Stecklov2009) note, chronic threat often triggers coping mechanisms such as emotional disengagement or denial of the reality, which can impair the response to political messages. Regulatory fit theory (Higgins, Reference Higgins1998; Mannetti et al., Reference Mannetti, Brizi, Giacomantonio and Higgins2013) also suggests that in high-threat contexts, emotionally saturated individuals may prefer messages that offer symbolic distance, helping to explain the relative effectiveness of the general framing over the localized one.
11. Discussion and conclusion
Can democratic promotion interventions foster public support for democracy in terrorism-affected contexts? Our findings suggest they can. A key outcome is that the interventions successfully reshaped public perceptions of democracy as a secure and effective system for combating terrorism. This highlights the potential of concise online videos presenting accurate empirical evidence to counter the perception of an “autocratic advantage” in conflict resolution (Merolla and Zechmeister, Reference Merolla and Zechmeister2019). By challenging the security-based narratives employed by autocratic leaders to consolidate power (Applebaum, Reference Applebaum2024), this approach underscores the instrumental value of democracy. Beyond its intrinsic virtues, democracy’s capacity to address urgent societal challenges and needs is crucial for persuading citizens facing insecurity and adversity (Higgins, Reference Higgins1998).
The study further examines various framings of democratic promotion interventions, from general approaches to context-specific ones. Among the interventions tested, the treatment emphasizing the universal advantages of democratic systems in addressing terrorism had the most significant impact, outperforming the treatment focused on universal rights. In contrast, the context-specific treatment, designed to address the direct experiences of the targeted population, proved ineffective. While this contradicts our initial expectation in hypothesis (H2b), which anticipated stronger effects from interventions tailored to the unique challenges faced by communities affected by terrorism, the result is consistent with hypothesis H2c and the broader literature on coping under chronic threat (Vázquez et al., Reference Vázquez, Pérez-Sales and Matt2006; Spilerman and Stecklov, Reference Spilerman and Stecklov2009).
This finding must be interpreted in light of the broader context in which the study took place. Just one month before data collection, a terrorist attack claimed the lives of more than 600 people,Footnote 23 the highest single-day death toll in the country’s history. Beyond this acute tragedy, the data reflect a broader pattern of entrenched insecurity: 77% of respondents reported direct or indirect exposure to terrorism, including injury, displacement, or the loss of a close relative. These figures are not isolated; they reflect the cumulative impact of nearly a decade of sustained terrorist violence. The country has transitioned from episodic attacks to a state of chronic insecurity, where exposure to violence, both physical and psychological, has become widespread, personal, and enduring.
Despite this backdrop, our results show no evidence that emotional states (anger, anxiety) or proximity to attacks moderated the impact of democratic promotion messages. These null results challenge the assumption that heightened threat salience intensifies political receptiveness. Instead, they align with research in Israel and Northern Ireland suggesting that prolonged exposure produces desensitization and psychological disengagement (Spilerman and Stecklov, Reference Spilerman and Stecklov2009). Although Burkina Faso’s experience with terrorism is more recent, its intensity and frequency since 2015 make desensitization highly relevant.
The concept of dissociative coping, well-documented in Northern Ireland, is particularly relevant (Spilerman and Stecklov, Reference Spilerman and Stecklov2009). In high-threat environments, people may resort to psychological defenses such as denial, detachment, and blame projection to maintain a sense of normalcy. These defenses reduce distress but can also diminish the credibility of political messages. Our manipulation checks reinforce this view: both instrumental democratic interventions—and especially the Burkina Faso-specific video—were less likely to be seen as informative, relatable, or shareable than the placebo. Far from being more persuasive, the contextualized message may have been perceived as inauthentic, politicized, or emotionally taxing.
The absence of a “low point” or an end to the violence means that elevated fear or anxiety becomes a permanent feature, not a transient response (Spilerman and Stecklov, Reference Spilerman and Stecklov2009). In this context, abstract democratic framings may hold more symbolic appeal than locally grounded but discredited narratives tied to politics and broken promises.
These findings suggest that in contexts of chronic insecurity, democracy promotion must move beyond assumptions of emotion-based receptiveness and attend instead to deep-seated trust deficits and narrative fatigue. Interventions that evoke hope, dignity, or international solidarity, rather than reiterating local trauma, may stand a better chance of being heard.
These results offer practical implications for democratic promotion in fragile and authoritarian settings. First, they suggest that context-sensitive messaging must be designed with great caution in high-threat environments. Although intuitive, localized appeals may underperform or even backfire if they intersect with trauma, distrust, or state-controlled narratives. Second, short, low-cost online interventions, such as those deployed in this study, can meaningfully shape public attitudes toward democracy when framed appropriately. They are particularly valuable in contexts where in-person programming is dangerous or infeasible. Social media can spread autocratic narratives, yet research shows that short-video interventions can sustain effects for at least two weeks and hold significant potential for democratization (Neundorf et al., Reference Neundorf, öztürk, Finkel and Ramírez2025).
At the same time, structural challenges remain. Authoritarian regimes often limit access to media and information, and online platforms are increasingly subject to surveillance or censorship (Analytica, Reference Analytica2018). Nevertheless, online advertisements and video content continue to offer outreach opportunities, especially when targeting platforms beyond the reach of domestic authorities. Such tools cannot replace long-term democratic development, but they may help maintain civic engagement in hostile settings.
This study raises important questions for future research. In contexts of chronic terrorism, under what conditions do citizens remain open to democratic appeals? When does the desire for security override habituation to violence? Further work is needed to examine the specific types of narratives that generate trust and engagement, and how these dynamics vary across countries and conflict stages. Understanding how psychological fatigue, institutional context, and narrative design interact will be essential for developing more effective democratic interventions in settings of sustained political violence.
Finally, we hope that future research will also advance the generalizability of our findings. One important limitation of this study is the coverage bias in our sample. Facebook sampling enabled us to reach broader and more diverse populations than would otherwise have been feasible in Burkina Faso. Nevertheless, our sample still over-represents more educated individuals. Weighting and conditional effects analyses across education categories did not reveal significant biases, which alleviate some concerns. Still, it is possible that segments of the population not captured at all, such as those without internet access or formal education, might respond differently. Replicating the study with fully representative samples and using survey designs that trace the long-term effects of repeated exposure to online content would be valuable steps forward. As digital infrastructures and recruitment methods continue to develop, future research will be able to draw on more inclusive and reliable strategies that further reduce these limitations.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/psrm.2025.10075. To obtain replication material for this article, https://doi.org/10.7910/DVN/I6DT0K
Acknowledgements
We thank Steven Finkel, Ericka Rascon Ramirez, Sergi Pardos Prado, Bernhard Reinsberg, Fred Paxton, and the participants of the Comparative Politics research cluster at the University of Glasgow for their valuable feedback on a previous draft. The authors made use of ChatGPT (GPT-5, OpenAI) to assist with English language editing only. The tool was accessed via ChatGPT (https:/chat.openai.com) without modification. All analysis, drafting, and interpretation are the sole responsibility of the authors.
Funding statement
The research of this project is generously funded by a European Research Council Consolidator Grant “Democracy under Threat: How Education can Save it” (DEMED) (Grant number: 865305).
Preprint (OSF)
The preprint can be accessed at this link: https://osf.io/preprints/osf/b4hju_v1.
Competing interests
The author(s) declare none.
Ethical statements
The study received ethical approval (number: 400230234) from the Ethics Committee of the College of Social Sciences of the University of Glasgow on 26 July 2024. After we received ethical approval and registered our research design, we collected data between 2 November and 3 December 2024. The registration can be accessed through this link: https://osf.io/5rweh/overview.
Open access
