Political Polarisation

A growing body of research tests interventions aimed at reducing political polarisation or mitigating its effects on how citizens perceive, interpret, and engage with political information. These interventions build on the mechanisms identified in the previous sections – social cleavages, elite strategies, emotional drivers, and digital amplification – and offer diverse entry points for reform. While most studies focus on the role of information and media exposure, recent work spans a broader set of strategies, including elite rhetoric, institutional reforms, social contact, and psychological reframing. Table 2 summarises the major intervention categories.

Table 2: Summary of Polarisation Interventions

Note: This table summarises major intervention types by the categories used in the paper, including experimental and observational evidence on effectiveness.

Social Contact and Dialogue Interventions

Interventions in this category focus on increasing constructive contact between partisan or identity groups to reduce prejudice and promote mutual understanding. Evidence shows that meaningful cross-group interactions – especially those emphasising shared identities or common experiences – can effectively reduce affective polarisation.

Some interventions focus on fostering positive dialogue and contact between individuals across the ideological aisle (Hartman et al. 2022). Experimental work in diverse contexts, from online deliberation platforms in the US (Fishkin et al. 2021, Levendusky and Stecula 2023, Combs et al. 2023) to in-person programmes in developing countries, suggests that structured interaction across group lines can significantly reduce hostility and increase empathy (Greene et al. 2024, Lowe 2021). For instance, Lowe (2021) shows that mixed-caste cricket teams in India significantly reduced caste-based biases, while Ghosh et al. (2025) demonstrate that collective rituals, such as singing the national anthem and wearing uniforms, can reduce in-group bias and increase willingness to interact with out-group members, particularly among the Hindu majority in youth camps.

Other interventions emphasise shared identities and cross-cutting affiliations as a way to build social contact across ideological divides. For example, Levendusky (2018) finds that highlighting a shared American identity reduces affective polarisation between Republicans and Democrats. Similarly, Ajzenman et al. (2023) shows that shared sports fandom increases the likelihood that users will follow cross-partisan accounts on social media. These results reinforce the idea that even modest signals of commonality can disrupt negative partisan stereotypes and encourage openness.

Building on this, Voelkel et al. (2024) conducted a large-scale field experiment with over 35,000 participants in the US to evaluate the effectiveness of more than 25 anti-polarisation interventions. They find that approaches emphasising shared identities and common experiences across partisan lines are among the most effective at reducing affective polarisation – highlighting the power of social contact framing even in low-touch or virtual formats.[1]1. These findings straddle both interpersonal (contact-based) and intrapersonal (cognitive-affective) mechanisms, illustrating how relational and emotional framing can jointly reduce partisan hostility.  Importantly, the creation of a new common identity can redefine the boundaries of in-groups and out-groups, potentially generating hostility towards another out-group (Fouka et al. 2021, Fouka and Tabellini 2022).

Elite-driven and Institutional Interventions

As noted in the previous sections, polarisation is not only a grassroots phenomenon. Political elites and institutions play a central role in shaping how divisions emerge, evolve, and are resolved. As such, a growing body of research explores interventions that target polarisation from the top down: through changes in elite messaging and institutional design.

Experimental evidence suggests that elite rhetoric can meaningfully influence citizens’ attitudes and behaviours. For example, Weiss et al. (2025) find that bipartisan television ads in the US increased openness to cross-party conversations and willingness to engage in bipartisan behaviour, though they did not significantly reduce affective polarisation. More generally, when elites model respectful engagement across divides, they may help reset partisan norms and expand the perceived boundaries of legitimate disagreement.

In less developed democracies, recent work explores whether party campaign strategies can reduce polarisation among voters. Cruz et al. (2024) conduct a randomised experiment during the 2019 Philippine elections, testing a set of non-partisan informational messages and emotionally framed appeals. Information-based messaging had more consistent depolarising effects, particularly among opposition voters. These findings suggest that cues from candidates and political parties can shape political attitudes even in highly polarised settings, but their effectiveness hinges on message type and partisan alignment.

Beyond rhetoric, structural reforms that alter the rules of political competition may also help mitigate polarisation. For example, Grose (2020) shows that California’s top-two primary system moderates legislative behaviour by incentivising broader appeals to the electorate. Similarly, Peskowitz and Szewczyk (2022) find that proportional electoral systems are associated with lower legislative polarisation than majoritarian ones, likely because they reduce the zero-sum nature of political competition.

While more causal evidence is needed – especially outside the US – these studies point to promising institutional levers for reducing elite-driven polarisation and its downstream effects on citizens.

Psychological and Cognitive Interventions

Psychological and cognitive interventions aim to reduce polarisation by targeting how individuals interpret and evaluate political information – especially when that information challenges their prior beliefs. These approaches do not simply seek to provide factual corrections; rather, they attempt to reshape the cognitive and emotional filters through which political messages are processed.

Many rely on subtle cognitive ‘nudges’ to reduce biased reasoning and increase receptivity to counter-attitudinal information. For example, Bolsen and Druckman (2015) use survey experimental evidence to show that warning against politicisation can counteract political effects when processing counter-attitudinal information.

Other ‘nudges’ work by activating positive dispositions. For example, making the value of open-mindedness salient has been shown to reduce partisan-motivated reasoning and increase acceptance of ideologically inconsistent facts (Groenendyk and Krupnikov 2021). Baron (2019) further suggests that priming open-mindedness affects information processing by shaping which sources individuals deem credible – operating on the upstream heuristics of trust in information.

Recent field experiments confirm the promise of these approaches beyond lab settings. In the politically polarised context of Mexico, Enriquez et al. (2025) show that priming voters not to react emotionally or politicise reverses the backlash typically triggered by providing counter-attitudinal performance information about incumbents. Larreguy and Tiburcio (2025) also find that similar nudges reduce motivated processing of counter-attitudinal news.

These interventions appear most effective when they operate at the meta-cognitive level: not aiming to replace beliefs, but to shift how beliefs are evaluated. By fostering critical reflection, dampening defensiveness, and loosening the grip of partisan identity on reasoning, they offer scalable and relatively low-cost tools for depolarisation. However, their impact likely depends on contextual factors such as political salience, social identity salience, and baseline institutional trust, highlighting the need for more research on when and where these tools are most effective.

Information Environment and Traditional Media

A major class of interventions seeks to reshape the information environment by altering media consumption patterns, moderating content, and correcting misinformation. These strategies aim to counter polarisation by improving access to credible information and reducing ideological misperceptions.

Another body of work evaluates fact-checking and corrective information campaigns. These interventions aim to reduce polarisation by correcting false beliefs – especially misperceptions about political opponents. Timely, credible corrections can temper extreme attitudes, particularly when framed in a non-partisan way (Guess et al. 2020, Lazer et al. 2018). However, their effectiveness hinges on audience trust and political context.

Other interventions focus more directly on correcting misinformation about political out-groups. In the US, Ahler and Sood (2018), Lees and Cikara (2019), and Druckman et al. (2022) show that correcting exaggerated or false beliefs about political opponents can significantly reduce affective polarisation. Even when belief change occurs, behavioural indicators (e.g. media consumption, cross-party engagement) may remain unchanged unless trust in media also increases.

Evidence from sub-Saharan Africa underscores how similar dynamics can operate in ethnically polarised environments. While they are not framed in terms of polarisation, studies in Ghana and Côte d'Ivoire show that balanced or non-partisan media can reduce ethnically charged political attitudes (Adida et al. 2020, Conroy-Krutz and Moehler 2015). In Sierra Leone, Casey (2015) finds that access to community radio increased citizens' political knowledge and led them to vote based on candidates' proposed policies rather than their ethnic backgrounds. Though these studies do not directly measure affective polarisation, they suggest that trusted, balanced media can promote ideological moderation and cross-group understanding in multi-ethnic, low-trust settings.

Taken together, these findings suggest that success depends not just on exposure but also on credibility, sustained engagement, and group identity salience. Interventions that build media trust and deliver content through trusted messengers may be more effective in polarised or divided contexts.

Social Media and Algorithmic Interventions

A related set of interventions targets the structure of online platforms, where algorithmic curation amplifies emotionally charged, ideologically aligned content. Several studies test whether disrupting these algorithmic dynamics reduces polarisation.

One approach focuses on increasing exposure to counter-attitudinal or balanced news sources. In the US, Levy (2021) finds that encouraging Facebook users to follow cross-partisan outlets increases consumption of those outlets and reduces affective polarisation, though not issue-based polarisation. By contrast, Broockman and Kalla (2022) show that sustained exposure to CNN among Fox News viewers moderates policy attitudes, even if it has no effect on affective polarisation. These findings highlight both the promise and limitations of media exposure interventions when affective divisions are entrenched.

Bail et al. (2018) find that exposing Twitter users to counter-attitudinal content via bots increased rather than reduced polarisation, likely due to backlash effects. However, more organic forms of exposure – such as encouraging users to follow real, cross-partisan news outlets – appear more effective, as discussed above.

Akbiyik et al. (2025) provide more nuanced evidence from the Turkish context. They show that while short-term exposure to ideologically distant news on Facebook and Twitter can initially backfire, sustained exposure over time reduces ideological polarisation, particularly when the ideological distance between users and the media source is smaller. In turn, while there are short-term reductions in affective polarisation, they tend to dissipate over time.

These studies suggest that social media interventions may require a dual approach: reducing misinformation and simultaneously increasing media trust. Repeated, voluntary exposure to credible, ideologically diverse sources – rather than one-time corrections or artificial bots – appears more promising for building long-term resilience to polarisation online.

While polarisation interventions vary widely in design and setting, several themes emerge. First, durable impact often requires sustained engagement, whether through repeated exposure, ongoing interaction, or long-term reforms. Second, interventions must match context: strategies that succeed in high-trust environments may be less effective or even backfire in low-trust or divided settings. Finally, emotional and cognitive cues matter as much as information. Effective polarisation interventions often aim not just to change what people know, but how they reason, relate to others, and emotionally respond to political information.

For full reference list see the end of the conclusion chapter.