Countering Polarisation: which interventions are most effective?
Polarisation is now widely recognised as a systemic risk to democratic stability, institutional trust, and social cohesion. Governments and practitioners are increasingly investing in programmes to address it—yet results remain mixed.
Polarisation is no longer an abstract concern—it is something many practitioners encounter directly in their work. Whether in communities, workplaces, education, or civic programmes, rising mistrust and hostility between groups are making engagement harder, collaboration more fragile, and outcomes less predictable.
In response, a wide range of interventions have emerged: dialogue programmes, media campaigns, training modules, and community initiatives. Yet many practitioners report a common frustration: what works in theory often has limited or inconsistent impact in practice.
The research helps explain why. There are evidence-based approaches to reducing polarisation—but they depend heavily on design, sequencing, and context. This guide translates that evidence into practical insights for those designing and delivering interventions on the ground.
The Evidence Base: Modest Effects, Strong Signals
Across multiple meta-analyses and experimental studies, interventions aimed at reducing affective polarisation—hostility toward opposing groups—tend to produce small but statistically significant effects. Typical shifts are in the range of a few points on feeling thermometers or attitudinal scales, and these effects often decay over time (e.g. Druckman et al., 2021; Paluck et al., 2021) . Two implications follow.
First, expectations must be realistic. No single intervention is likely to produce large or lasting change on its own. Second, these small effects are still meaningful at scale—particularly when applied repeatedly across populations and institutions.
The evidence also shows high design sensitivity. Small changes in framing, messenger, or context can significantly alter outcomes. In some cases, poorly designed interventions can reinforce divisions rather than reduce them (e.g. Bail et al., 2018) .
Reframing the Goal: Reducing Hostility, Not Eliminating Disagreement
A key insight from the literature is that polarisation is not primarily about disagreement over issues. Democracies can sustain high levels of disagreement. What undermines them is affective polarisation—the tendency to distrust, dislike, and dehumanise opposing groups (Iyengar et al., 2019).
Interventions should therefore not aim to produce consensus. Instead, the goal is to reduce misperceptions about out-groups, lower perceived threat and increase willingness to engage across differences. The objective is to make disagreement less socially and politically toxic, not to eliminate it .
Three Evidence-Based Mechanisms
The literature converges on three mechanisms that consistently reduce affective polarisation: Correcting Misperceptions, Humanising the Outgroup, and Structured Intergroup Contact. These mechanisms operate through distinct but complementary psychological pathways.
1. Correcting Misperceptions and Social Norms
The evidence shows that individuals systematically overestimate the extremity and hostility of both opposing groups and, importantly, their own group. This phenomenon—sometimes referred to as pluralistic ignorance—creates a distorted perception of social norms (Prentice & Miller, 1993; Levendusky & Malhotra, 2016).
Recent studies demonstrate that providing accurate information about group attitudes can significantly reduce polarisation. For example Levendusky (2018) finds that correcting misperceptions about partisan hostility reduces animosity, Ahler (2014) shows that people overestimate how ideologically extreme their co-partisans are and Mernyk et al. (2022) demonstrate that correcting “false polarization” reduces support for partisan hostility. These findings align with the broader evidence base: misperceptions are a key driver of division, and correcting them reduces perceived threat.
Misperceptions amplify conflict through two mechanisms. Threat inflation — people believe the other side is more extreme than it is. Norm distortion — people believe hostility is more socially acceptable than it is. Correcting these beliefs reduces both fear and the perceived legitimacy of antagonistic behaviour.
Norm-correction interventions are relatively low-cost, highly scalable and suitable for mass communication. Effective formats include public messaging campaigns highlighting shared values, internal organisational communications (e.g. in workplaces or universities), and data-driven storytelling showing actual opinion distributions.
Crucially, these interventions are most effective when delivered by trusted messengers and framed in non-partisan ways.
2. Humanising Through Narrative
The evidence shows that narrative-based interventions—particularly those that present personal stories from members of out-groups—can reduce prejudice and dehumanisation. This builds on a large body of research, Broockman & Kalla (2016) show that brief, structured conversations using personal narratives can durably reduce prejudice (in their case, toward transgender people), Paluck & Green (2009) highlight the role of media narratives in shifting social norms, Vezzali et al. (2014) demonstrate that even indirect or imagined contact through stories can reduce intergroup bias.
Narratives work not by changing beliefs about policy, but by changing how individuals perceive the people behind those positions. Narratives activate several mechanisms: Individuation — replacing abstract group stereotypes with concrete individuals, Perspective-taking — increasing cognitive empathy, and Identity complexity — highlighting overlapping identities beyond political affiliation. These processes weaken “us vs. them” thinking and reduce dehumanisation.
Effective narrative interventions prioritise personal, relatable stories over political argument, highlight shared experiences or identities (e.g. parenthood, work, place) and avoid overtly partisan framing. Delivery channels matter. Evidence suggests higher impact when narratives are locally grounded (e.g. community media), repeated over time (e.g. serial storytelling), embedded in trusted institutions.
For policymakers, this points to partnerships with local media, civic organisations, and public institutions as key delivery mechanisms.
3. Structured Intergroup Contact
The evidence shows that intergroup contact is one of the most studied interventions in social psychology. The foundational contact hypothesis (Allport, 1954) has been extensively validated and refined. Meta-analyses (e.g. Pettigrew & Tropp, 2006) show that contact generally reduces prejudice—but only under certain conditions: equal status between groups, shared goals, cooperation rather than competition and institutional support. More recent work highlights that contact can fail—or backfire—if these conditions are not met, particularly in highly polarised contexts (Enos, 2014) .
Structured contact reduces polarisation through: anxiety reduction — lowering fear of the out-group, trust-building — enabling positive expectations of interaction, and norm change — demonstrating that cooperation is possible and acceptable.
Unlike information or narrative interventions, contact creates direct relational experience, which can produce deeper and more durable change. Contact interventions are high-impact but resource-intensive, less scalable than messaging or narrative approaches and highly sensitive to design quality. They are most effective when participants work toward shared, practical goals, engagement is repeated rather than one-off, and facilitation is skilled and neutral. Examples include community problem-solving groups, workplace team-based collaboration across differences, deliberative forums with structured facilitation.
Sequencing: Why Order Determines Effectiveness
One of the most important operational insights is that these interventions are not interchangeable. Their effectiveness depends on sequencing and the evidence supports a three-stage model:
Norm correction reduces perceived threat
Narrative exposure builds empathy
Structured contact enables constructive interaction
This sequence matters because individuals are unlikely to engage productively in contact if they perceive the other group as hostile or extreme. Early-stage interventions reduce these barriers and increase the likelihood that contact will succeed. For policymakers, this implies that launching dialogue programmes without prior norm and narrative work is likely to produce weak or inconsistent results.
From Interventions to Systems: Where This Works in Practice
The evidence consistently points to the importance of institutional context. Interventions are most effective when embedded in environments that provide repeated exposure, trusted authority or legitimacy, and opportunities for sustained interaction.
This makes certain settings particularly promising
Workplaces: enable repeated, cooperative interaction and integration into training and culture
Educational institutions: provide structured, long-term engagement
Local government: leverages place-based identity and trusted convening power
Civic organisations: support recruitment, legitimacy, and facilitation
Local media: deliver trusted, scalable narrative interventions
These environments allow interventions to move from isolated events to ongoing social processes.
Practical Principles for Effective Design
Drawing on the evidence, six principles should guide implementation:
Design for repetition, not one-off impact, as effects decay; hence reinforcement is essential
Prioritise mechanism over format. The focus should be on what psychological process is being activated
Sequence interventions deliberately. Norms → narratives → contact
Use trusted messengers and institutions. Credibility is a key determinant of effectiveness
Avoid adversarial or politicised framing. Emphasise shared norms and identities
Test, iterate, and evaluate. Small design changes can significantly affect outcomes
Conclusion: Building the Conditions for Less Toxic Disagreement
The strongest insight from the evidence is that polarisation is not just an individual attitude problem—it is shaped by social environments. Polarisation is sustained by distorted perceptions, dehumanising narratives, and limited cross-group interaction. The best forms of countering it come from shaping those environments: what people think others believe, what stories they encounter, who they interact with, and how.
No single intervention will solve polarisation. But well-designed, repeated, and sequenced interventions—embedded in everyday settings—can meaningfully reduce hostility over time and rebuild the foundations of constructive disagreement. That is both a more realistic goal and a more achievable one.
For policymakers and practitioners, the task is not simply at a programme level, but to build systems that continuously reinforce these dynamics across institutions. The goal is not consensus, it is a society in which disagreement is possible without mutual distrust—and that is both achievable and essential for democratic resilience.
