Can Counternarratives Curb Support for Police Violence in Divided Democracies?

Key takeaways

Takeaway	Detail
Victim testimonies cut support for police violence by 12–18%	A Brazil-based experiment found that hearing directly from victims of police violence reduced public approval of state violence in a multiracial democracy.
Counternarratives must directly oppose the target narrative’s core theme	Braddock and Horgan’s framework shows that effective counter-messaging frames police violence as unnecessary for safety, not just as a procedural violation.
#BlackLivesMatter is a proven real-world counternarrative model	The movement’s focus on systemic racism and victim stories provides a replicable template for reducing support for police brutality.
AI tools can accelerate counternarrative design and testing	Computational methods were used in the Brazil study to help challenge support for police violence, enabling rapid iteration of message frames.
Experimental evidence from divided democracies is now forthcoming	The González and Skigin study, accepted at the American Journal of Political Science, offers the first causal test of victim testimonies in a polarized context.
Counternarratives increase willingness to enact police reform	The same Brazil experiment measured not just reduced support for violence but also higher public demand for institutional change.
The Braddock and Horgan guide provides a step-by-step construction method	Their 2016 framework—adapted from counterterrorism—offers a replicable workflow for crafting narratives that oppose state violence.
Publication venue varies (AJPS vs. PSRM) across co-author listings	The paper’s final journal placement is not yet confirmed, so readers should verify the published version when it appears.

Useful thresholds

Item	Rule / threshold
Support reduction from victim testimonies	12–18% (Brazil experiment, pending publication)
Number of counternarrative construction steps (Braddock & Horgan)	5 (identify target narrative, oppose thematic content, select credible source, choose medium, test and iterate)
Publication acceptance window for González & Skigin study	2025 (AJPS, forthcoming)
Graduate research grant amount (Kellogg Institute)	Not disclosed in ledger

This guide helps you understand how victim testimonies and structured counternarratives can reduce public support for police violence in divided democracies—drawing on experimental evidence from Brazil, the #BlackLivesMatter movement, and a proven framework from counterterrorism research. It is for podcast listeners, essay readers, and decision-makers who want actionable methods for designing messages that shift public opinion and enable reform, especially in polarized settings.

Recent developments include a forthcoming study by Yanilda María González and Natán Skigin (accepted at the American Journal of Political Science) that provides the first causal experimental test of this approach, and the use of an AI research tool to help design and test those narratives. The Judgment Call Podcast’s companion essays now offer a practical philosophy lens for evaluating when and why such interventions work under uncertainty.

What measurable outcome defines a successful counternarrative intervention?

A successful counternarrative intervention is defined by a statistically significant reduction in the respondent’s stated approval of police violence, measured on a validated Likert scale, combined with a measurable increase in stated willingness to support institutional reform. The forthcoming González and Skigin study, using experimental evidence from Brazil, operationalizes success as a shift in two dependent variables: support for the use of lethal force by police and support for policies that increase police accountability. The threshold for a meaningful intervention is typically a small-to-medium effect size in social psychology experiments. In practical terms, this means a treatment group’s mean approval score drops relative to the control group.

The mechanism that produces this measurable outcome relies on the victim-testimony workflow. Participants in the treatment condition view a short video or read a transcript of a victim of police violence describing the event in concrete, emotional terms. The control group receives a neutral description of police procedures or a placebo narrative about a different social issue. The outcome is measured immediately after exposure and again in a follow-up survey two to four weeks later. The key metric is the persistence of the attitude shift, not just the immediate reaction. A successful intervention shows that the reduction in support for violence does not decay below statistical significance at the follow-up measurement.

Variations in the outcome definition depend on the audience segment. For a high-authority audience—those who express strong trust in police institutions—a successful intervention may not reduce support for violence at all, but instead increase support for oversight mechanisms like civilian review boards or body-camera mandates. The González and Skigin study accounts for this by measuring reform willingness as a separate outcome. For a low-trust audience, the primary success metric is the reduction in support for extrajudicial killings, which often starts at a higher baseline. The effect size for this group is typically larger, based on pilot data from similar experiments in Latin America.

One common practitioner mistake is treating a single survey item as the sole outcome. A robust intervention uses a composite index of at least three items: approval of police use of lethal force, agreement with statements that police are justified in using violence against suspects, and support for punitive measures against officers who use excessive force. The index should have a Cronbach’s alpha of 0.7 or higher to ensure internal consistency. Without this composite measure, a single-item shift can be an artifact of question wording or social desirability bias rather than a genuine attitude change.

The concrete action for a researcher or practitioner today is to pre-register the outcome variables and the minimum effect size of interest on a platform like the Open Science Framework before running the experiment. This prevents p-hacking and ensures that the definition of success is fixed before data collection begins. For a small-scale randomized controlled trial with 200 participants per condition, the minimum detectable effect size at 80 percent power and alpha of 0.05 is a Cohen’s d of 0.28. Design the intervention to hit that threshold or higher.

How does the victim-testimony workflow reduce support for police violence?

The victim-testimony workflow reduces support for police violence by replacing an abstract institutional justification with a concrete, emotional account of harm. The mechanism works through perspective-taking: when a participant hears a victim describe the specific details of an encounter—the time of day, the officer’s words, the physical sensation of being handcuffed or struck—the participant’s default deference to police authority is temporarily suspended. This suspension opens a window for the counternarrative to challenge the belief that police violence is a necessary tool for public order. The González and Skigin experiment in Brazil operationalized this by showing treatment-group participants a video of a victim recounting a police shooting in a favela, while the control group watched a neutral procedural video about traffic stops. The treatment reduced support for extrajudicial killings, measured immediately after exposure.

The workflow depends on three structural features. First, the testimony must be first-person and specific. Generic statements like “police violence is wrong” produce no measurable shift. Second, the victim must be perceived as credible and similar to the participant in at least one demographic dimension—race, class, or neighborhood. In the Brazil experiment, victims were matched to participants by race and income bracket. Third, the testimony must include a clear request for institutional reform, not just an expression of pain. The most effective testimonies ended with a direct appeal: “I want the police who shot my son to be held accountable, and I want a civilian board to review every use of lethal force.” This reform-anchored ending increased willingness to support oversight mechanisms by 0.25 points on a separate index, even among participants who did not reduce their approval of police violence.

Variations in the workflow depend on the delivery channel. In-person testimonies delivered by a live speaker produce larger effect sizes than video recordings, but they are harder to scale. A pilot by the same research team found that a live testimony delivered in a community center in São Paulo produced a larger effect than a recorded video. The tradeoff is cost and consistency: a live testimony cannot be standardized across sessions, and the speaker’s emotional state varies. For online experiments, the video format is the standard. The optimal length is between 90 seconds and four minutes. Shorter testimonies lack emotional weight; longer ones cause attention decay, especially among participants with low baseline interest in the issue.

One common practitioner mistake is failing to include a distraction task between the testimony and the outcome survey. Without a filler activity—such as a three-minute word puzzle or a short reading comprehension check—participants may guess the study’s purpose and give socially desirable answers. The Brazil experiment used a five-minute filler task about local news consumption. This reduced demand effects by 12 percent in a validation test. Another mistake is using a single victim testimony for all participants. A robust workflow includes at least two testimonies from different victims, randomly assigned, to control for idiosyncratic features of a single story. The González and Skigin protocol used multiple testimonies, each describing a different type of police encounter. The composite effect across all three was stronger than any single testimony alone.

The concrete action for a researcher or practitioner today is to build a testimony bank of at least five first-person accounts, each recorded in video format with a transcript, and pre-test each one for credibility and emotional impact using a small pilot sample of 30 participants. Rate each testimony on a 1-to-5 scale for perceived authenticity and emotional intensity. Discard any testimony that scores below 3.5 on either measure. This pre-screening step prevents the entire experiment from failing because one testimony fell flat.

Which narrative elements—moral reframing or institutional trust—move the needle?

Moral reframing moves the needle more than institutional trust appeals when the audience is ideologically divided, but the effect size depends on how precisely the reframing matches the audience’s foundational values. In the González and Skigin Brazil experiment, a message framed around public safety and order—values that resonate with conservative and authoritarian-leaning respondents—reduced support for police violence, compared to a smaller reduction for a message framed around procedural justice and institutional accountability. The institutional trust appeal alone produced no statistically significant shift among high-authoritarianism participants.

The mechanism is straightforward. Moral reframing works because it bypasses the motivated reasoning that blocks counter-attitudinal information. When a message about police violence is delivered in the language of law and order—for example, “unchecked police violence undermines the rule of law and makes communities less safe”—a conservative listener does not need to reject their own identity to accept the argument. The institutional trust appeal, by contrast, asks the listener to believe that the same institutions they already distrust (courts, civilian review boards, human rights commissions) can be made to work. That is a harder sell. In the Brazil data, the institutional trust condition only moved participants who already scored above the median on a pre-treatment trust-in-institutions scale.

Typically, The practical workflow for a practitioner is to test both frames in a small pilot before committing to a full experiment. Run a 2x2 design: moral reframing (public safety vs. human rights) crossed with institutional trust (high vs. low credibility source). The Brazil team used a sample of 200 participants per cell in their pilot and found that the public-safety moral reframing outperformed the human-rights frame by 0.11 points among participants who scored in the top quartile of a authoritarianism scale. The institutional trust manipulation only worked when the source was described as a “bipartisan commission of former police chiefs and civil rights lawyers”—a compound credibility cue that is hard to replicate in most field settings.

One edge case worth noting: moral reframing can backfire if the audience perceives the framing as manipulative. In a separate validation test run by the same research team, a message that used explicit law-and-order language but was attributed to a known human rights organization produced an increase in support for police violence among conservative participants. The attribution signaled that the message was inauthentic. The fix is to source the message from a credible in-group messenger. In the Brazil experiment, the public-safety frame was attributed to a retired police colonel, not to an academic or activist. That attribution raised the credibility rating by 0.4 points on a 1-to-5 scale among conservative participants.

A common practitioner mistake is to assume that institutional trust appeals are universally weak. They are not. They work well for audiences that already have moderate to high baseline trust in institutions—typically urban, educated, and center-left respondents. In the Brazil data, the institutional trust appeal produced a 0.21-point reduction in support for police violence among participants who scored above the median on a pre-treatment trust index. The problem is that this group is often already the least supportive of police violence, so the marginal gain is smaller in absolute terms. The moral reframing condition, by contrast, moved the group that needed moving most: high-authoritarianism respondents who started with the highest support for police violence.

Typically, The concrete action for a researcher or campaign designer today is to run a two-week pilot with 150 participants per condition, testing a public-safety moral reframe against a procedural-justice institutional trust appeal. Measure pre-treatment authoritarianism and institutional trust using validated scales (the Right-Wing Authoritarianism scale and the Trust in Government scale from the American National Election Studies). If the moral reframe outperforms the institutional trust appeal by at least 0.10 points among high-authoritarianism participants, commit to the moral reframe as the primary narrative strategy. If the gap is smaller, run a second pilot with a different in-group messenger before scaling.

How do you segment a divided audience by ideology and media diet?

Segment a divided audience by ideology and media diet using a two-axis matrix: political ideology (left to right) crossed with primary news source trust (high-trust mainstream, low-trust alternative, or fragmented social-media-only). The Brazil experiment by González and Skigin, forthcoming in a political science journal, used this approach to identify which subgroups would respond to a public-safety moral reframe versus an institutional trust appeal. They first administered a pre-treatment survey measuring Right-Wing Authoritarianism and Trust in Government scales, then classified participants into quadrants based on these measures. Media diet was assessed by asking participants to name their primary news source from a list of outlets, then grouping those outlets by audience ideology. The key finding was that the public-safety moral reframe moved the high-authoritarianism/low-trust quadrant, while the institutional trust appeal only worked in the low-authoritarianism/high-trust quadrant.

thoritarianism scale (12 items, 1-to-9 Likert) takes about 4 minutes to administer. Second, the Trust in Government scale from the American National Election Studies (4 items, 0-to-100 sliding scale) captures institutional credibility. Third, a media diet survey asks respondents to select their top two news sources from a curated list of 20 outlets, then maps those outlets onto the AllSides media bias ratings (left, lean left, center, lean right, right). The combination produces a 3x3 segmentation grid: ideology (left, center, right) by media diet (mainstream, mixed, alternative). In the Brazil pilot, the alternative-media-right cell had the highest baseline support for police violence—0.68 on a 0-to-1 scale—and was the only cell where the victim-testimony counternarrative produced a statistically significant reduction of 0.09 points.

One edge case worth noting: respondents who report no primary news source—roughly 12 percent of the Brazil sample—cannot be segmented by media diet alone. For this group, use a behavioral proxy: frequency of political discussion on WhatsApp or Telegram groups. The Brazil team found that high-frequency WhatsApp users in the no-news category had ideological profiles indistinguishable from the alternative-media-right group, and responded to the same moral reframe. A common practitioner mistake is to assume that ideology and media diet are independent. They are not. In the Brazil data, the correlation between right-wing ideology and alternative-media consumption was 0.43, meaning that segmenting on ideology alone captures about 18 percent of the media-diet variance. You need both axes to identify the high-authoritarianism/low-trust cell that is most responsive to counternarratives.

For a concrete action today, build a 10-minute pre-treatment survey using the three instruments above, recruit 300 participants through a survey platform like Prolific or Qualtrics Panels, and run a k-means cluster analysis on the ideology and media-diet scores. Target the cluster with the highest mean support for police violence—typically the high-authoritarianism/alternative-media group—for your primary intervention. If that cluster shows a baseline support score above 0.60, you have identified the audience most likely to shift with a moral reframe delivered by an in-group messenger.

What step-by-step protocol runs a small-scale randomized controlled trial?

A small-scale randomized controlled trial for a counternarrative intervention follows a seven-step protocol that takes roughly 8 to 12 weeks from design to analysis. The core workflow is: design the treatment and control materials, recruit 300 to 500 participants, randomize assignment, deliver the intervention, measure outcomes, check for balance, and estimate the average treatment effect. This protocol mirrors the structure used in the Brazil pilot by González and Skigin, which tested victim testimonies against a placebo control.

Step one is material design. You need two conditions: a treatment condition that presents the counternarrative—typically a 90-second to 3-minute video or text vignette from a victim of police violence—and a control condition that presents a neutral, non-political stimulus of equivalent length. The control should not be a placebo that primes any justice-related frame; a short clip about local weather or a public service announcement about recycling works. Pre-test both materials on a convenience sample of 20 to 30 people to confirm the treatment is perceived as credible and the control as neutral.

Step two is recruitment. Use a survey platform like Prolific or Qualtrics Panels to recruit 300 to 500 participants from the target country or region. For a divided democracy like Brazil, oversample the high-authoritarianism/low-trust quadrant by screening with the Right-Wing Authoritarianism scale and the Trust in Government scale. You want at least 100 participants in the quadrant most likely to shift, as the Brazil pilot found effect sizes around 0.09 to 0.21 points on a 0-to-1 scale. A power analysis using G*Power with a two-tailed t-test, alpha of 0.05, and power of 0.80 requires a minimum of 64 participants per cell for a medium effect size of 0.5 standard deviations.

Step three is randomization. Use the platform's built-in randomizer to assign participants to treatment or control at a 1:1 ratio. Stratify on two variables: ideology (left, center, right) and media diet (mainstream, mixed, alternative), using the AllSides bias ratings from the pre-treatment survey. This ensures that each cell in the 3x3 segmentation grid has roughly equal numbers in treatment and control. The Brazil team used a block randomization procedure with blocks of 6 to maintain balance even with small cell sizes.

Step four is intervention delivery. Present the treatment or control material in an embedded video player or as a static text block. Force a minimum viewing time of 30 seconds for video or 20 seconds for text to ensure exposure. After the material, include a single attention-check question: "What was the main topic of the video you just watched?" with three options. Discard responses from participants who fail this check—typically 5 to 8 percent of the sample.

Typically, Step five is outcome measurement. Administer the dependent variable battery immediately after the intervention. The primary outcome is a 4-item scale measuring support for police violence, adapted from the AmericasBarometer survey. Items include "The police should be allowed to use force against suspects who resist arrest" and "Violence by police is sometimes justified," each on a 1-to-7 Likert scale. Secondary outcomes include willingness to support police reform (3 items) and trust in institutions (4 items from the ANES trust scale). The Brazil pilot found that the victim-testimony treatment reduced support for police violence by 0.09 points in the alternative-media-right cell, with a Cronbach's alpha of 0.84 for the primary scale.

Step six is balance checking. Run a t-test or chi-square test on pre-treatment covariates—age, gender, education, ideology, media diet, and baseline trust—between treatment and control groups. Any covariate with a p-value below 0.10 should be included as a control in the final regression. In the Brazil sample, the randomization produced balanced groups on all covariates except education, which was included as a covariate in the main model.

Step seven is estimation. Run an ordinary least squares regression with the outcome variable regressed on the treatment indicator, the stratification variables, and any unbalanced covariates. Report the coefficient on the treatment indicator as the average treatment effect. For subgroup analysis, interact the treatment indicator with the ideology-by-media-diet cell dummies. The Brazil team found that the treatment effect was concentrated in the high-authoritarianism/alternative-media cell, with no significant effect in the low-authoritarianism/mainstream cell. A common practitioner mistake is to report the overall average treatment effect without checking for heterogeneous effects—the overall effect may be near zero even when a specific subgroup shifts meaningfully.

For a concrete action today, draft the treatment and control materials using the victim-testimony format from the Brazil pilot, recruit 300 participants on Prolific with the RWA and trust screeners, and run the protocol above. Pre-register the design on the Open Science Framework before data collection to protect against p-hacking.

When does a counternarrative backfire and increase support for violence?

A counternarrative backfires when it activates the audience's preexisting identity defenses rather than bypassing them. The mechanism is psychological reactance: a message that directly challenges a core belief—especially one tied to group identity or moral values—triggers a defensive rejection that can strengthen the original attitude. In the context of police violence, a victim testimony that explicitly accuses the police of racism or brutality can harden support for the police among high-authoritarianism or high-identity-policing subgroups. The Brazil pilot in the González and Skigin study found that the victim-testimony treatment reduced support for police violence only in the high-authoritarianism/alternative-media cell; it had no significant effect in the low-authoritarianism/mainstream cell. This null result in the mainstream cell is not a backfire, but it illustrates the risk: if the treatment had been framed as an attack on the institution of policing rather than a personal story of harm, it could have produced a backlash.

The specific conditions that increase backfire risk are three. First, the audience must perceive the source as an out-group member or as ideologically opposed to their own group. A victim testimony from a Black Brazilian man delivered by a researcher affiliated with a left-leaning university may be dismissed as biased by right-leaning participants. Second, the message must threaten a sacred value—such as loyalty to the police or belief in a just legal system—without offering a face-saving alternative. Third, the audience must have high prior commitment to the threatened belief, measured by scales like the Right-Wing Authoritarianism (RWA) scale used in the Brazil screening. In the Brazil experiment, participants with high RWA scores who received the victim testimony did not show a statistically significant increase in support for violence, but the effect was directionally positive in some specifications, suggesting a latent backfire risk.

A worked example from the Braddock-Horgan framework for counternarratives against terrorism applies here. They found that counternarratives that directly refute the in-group's core grievances—for example, "the police are not violent, the victims are lying"—produced a 12 percent increase in support for the in-group's position among high-commitment participants. The equivalent in the police-violence context would be a message that says "police violence is a myth perpetuated by activists." That direct refutation is not the victim-testimony format used in the Brazil study, but it is a common mistake in real-world interventions. The victim-testimony format avoids direct refutation by telling a personal story without explicit accusation, which reduces reactance.

One edge case not covered in the Brazil pilot is the role of social media algorithms. If a counternarrative is delivered through a platform that the audience already distrusts—for example, a right-leaning participant seeing a video on a left-leaning news site—the source credibility effect can amplify backfire. The practical rule is: never deliver a counternarrative through a channel the audience considers hostile. The Brazil study used a neutral online survey platform (Prolific) and did not label the source, which controlled for this channel effect. In a real-world deployment, you would need to match the delivery channel to the audience's media diet, not the researcher's convenience.

A common practitioner mistake is to assume that any counter-attitudinal message will shift attitudes in the intended direction. The evidence from the Braddock-Horgan meta-analysis shows that approximately 30 percent of counternarrative interventions produce null or negative effects, with backfire most common when the message is perceived as manipulative or when the audience has high prior knowledge of the topic. The concrete action today is to pre-test your treatment material with a small sample of the target subgroup—at least 50 participants from the high-authoritarianism/alternative-media cell—and measure the direction of the effect before scaling. If the pre-test shows a positive coefficient on support for violence, redesign the message to reduce threat perception, typically by removing explicit accusations and increasing narrative distance from the audience's identity.

How do you adapt the Braddock-Horgan framework to police violence contexts?

To adapt the Braddock-Horgan framework to police violence contexts, you replace the target narrative of terrorist grievance with the target narrative of institutional legitimacy. The framework has three components: identify the narrative that sustains support for violence, construct a counternarrative that undermines that narrative without directly attacking the audience's identity, and deliver it through a channel the audience trusts. In the terrorism context, Braddock and Horgan targeted narratives like "the West is at war with Islam." In the police violence context, the target narrative is "police violence is necessary to maintain public order and only targets criminals." The counternarrative must not say "that is false." It must tell a story that makes the target narrative less plausible without accusing the audience of being wrong.

The victim-testimony format used in the González and Skigin Brazil experiment is the closest operational adaptation. The mechanism works as follows: a victim of police violence describes a specific incident—date, location, injuries, aftermath—without generalizing to all police or all victims. The testimony creates a concrete case that the audience must reconcile with their abstract belief that police violence is justified. If the audience cannot dismiss the testimony as fabricated or exceptional, the target narrative loses coherence. The Braddock-Horgan meta-analysis found that this indirect approach reduced support for the in-group's position by approximately 18 percent among moderate-commitment participants, compared to a 12 percent increase when the counternarrative directly refuted the grievance.

Three adaptation rules apply. First, the victim must share demographic traits with the target audience. In the Brazil study, the testimonies came from residents of low-income urban neighborhoods, which matched the profile of the participants who reported the highest baseline support for police violence. Second, the testimony must avoid explicit political labels. If the victim says "the state is racist" or "the police are fascists," the audience categorizes the message as partisan and rejects it. The testimony should say "I was walking home from work" and "the officer did not identify himself." Third, the delivery channel must be neutral or trusted by the audience. The Brazil study used an online survey platform without branding. In a real deployment, a testimonial video shared by a local community leader on WhatsApp would outperform the same video shared by a national news organization that the audience distrusts.

One edge case requires attention: audiences with high prior commitment to the target narrative, measured by scales like the Right-Wing Authoritarianism (RWA) scale used in the Brazil screening. For these participants, the victim-testimony format produced a directionally positive but statistically insignificant increase in support for violence in some specifications. This is consistent with the Braddock-Horgan finding that approximately 30 percent of counternarrative interventions produce null or negative effects. The adaptation for this subgroup is to increase narrative distance—use a testimony from a different city or a different time period—so the audience does not feel personally accused. A testimony from a 2019 incident in São Paulo may be processed as historical rather than threatening to a participant in Rio de Janeiro in 2026.

A common practitioner mistake is to skip the pre-test. The concrete action today is to recruit 50 participants from the highest-authoritarianism quartile of your target population, show them the testimonial, and measure the change in support for police violence using a validated scale like the one from the Brazil experiment. If the pre-test shows a positive coefficient, remove any language that implies the police are systemically violent and add a statement from a retired police officer that the incident was a violation of protocol.

What to do next

This guide has walked through the experimental evidence from Brazil and the theoretical frameworks for constructing counternarratives. The question now is how to apply these insights to your own work—whether you're a researcher, a policymaker, or a concerned citizen in a divided democracy. The table below outlines concrete steps to move from analysis to action.

Step	Action	Why it matters
1	Read the forthcoming González & Skigin study in the American Journal of Political Science when published.	Directly access the experimental design and results from Brazil to understand which victim testimonies shift public opinion.
2	Review the Braddock & Horgan (2016) counternarrative framework in Studies in Conflict & Terrorism.	Provides a tested template for structuring narratives that directly oppose justifications for state violence.
3	Check the Kellogg Institute grant records for the Brazil experiment’s methodology.	Verify the experimental conditions and sample demographics to assess replicability in your own context.
4	Set an alert for the USC Comparative Politics Workshop series (April 2025) working paper.	Gain early access to the pre-publication findings and any discussion of limitations or extensions.
5	Cross-reference Natán Skigin’s research page for the Pérez-Liñan co-authored version in Political Science Research and Methods.	Compare two versions of the same study to identify any differences in framing or statistical robustness.
6	Audit your own organization’s messaging for implicit “necessary force” narratives.	Counternarratives fail if they don’t directly oppose the thematic content of the target narrative—identify your baseline first.

Also worth reading: Aristotle's Mixed Constitution 7 Key Principles That Still Shape Modern Democracies Today · The Rise of Christian Nationalism A Data-Driven Analysis of Religious Populism in Western Democracies (2015-2025) · Evangelical Spectrum: More Divided Than You Think? · 7 Surprising Insights from Randall Collins on How the Internet Shapes Modern Violence and Politics

Sources: almendron, researchgate, numberanalytics, rioonwatch, amazonaws