What to do about polarization? Evaluating the evidence base of recommendations for depolarization

Summary

This systematic literature review interrogates how social science scholarship moves from empirical findings about polarization to prescriptive recommendations for depolarization. Analyzing 89 peer-reviewed articles (2013–2023) under PRISMA guidelines, the authors find a striking disconnect: while the field has accumulated substantial evidence about polarization’s drivers, the recommendations scholars offer for depolarizing societies are rarely grounded in their own empirical results. Instead, these recommendations lean on implicit normative ideals — most prominently deliberative democracy. The paper calls for greater transparency about both the evidence base and the value commitments shaping scholarly prescriptions.

Key Contributions

• First systematic mapping distinguishing empirically supported factors, explicit recommendations, and underlying normative ideals in (de)polarization scholarship.
• Documents a major evidence-recommendation gap, exposing the role of implicit normative reasoning in shaping policy-relevant claims.
• Articulates two unresolved structuring debates: knowledge/education vs. culturally resonant communication, and reducing echo chambers vs. cautioning against cross-cutting exposure.
• Introduces a conceptual distinction between “mitigative preventive” and “progressive” approaches to depolarization.
• Proposes a reform agenda for the field emphasizing transparency, methodological diversification, interdisciplinary collaboration, and a shift toward studying depolarization itself.

Methods

A PRISMA 2020 systematic review (preregistered on OSF) of Web of Science results for polarization-related terms across communication, sociology, political science, psychology, economics, and general social science. From 7,994 hits, the top-cited relevant articles per year were retained, yielding 89 after full-text screening. The authors applied a mixed deductive–inductive coding scheme (Krippendorff’s α ≈ 0.87) focused particularly on discussion sections, combined with grounded-theory coding (open, axial, selective) for inductive themes.

Findings

• The field is overwhelmingly skewed toward polarization (82/89) rather than depolarization (1/89); ideological polarization dominates over affective.
• Methodologically narrow: experiments, surveys, and content analyses prevail; only one study examined face-to-face communication.
• Best-supported drivers of polarization: confirmation bias/motivated reasoning, selective exposure/echo chambers, one-sided media framing, and exposure to dissonant views.
• Depolarization factors are understudied (26 studies); knowledge/media literacy and culturally resonant communication received the most empirical support.
• Evidence supports contradictory claims about cross-cutting exposure — it can both increase and decrease polarization depending on conditions.
• Only 1 study based its depolarization recommendation on its own evidence; 41 made recommendations unsupported by their analyses.
• Deliberative democracy (often via Habermas) is the dominant — usually implicit — normative ideal.

Connections

This meta-level critique sits above the empirical contributions in the topic cluster, putting their recommendation-generating practices under scrutiny. It speaks directly to debates about echo chambers and cross-cutting exposure exemplified by Bakshy2015-rn, and to recent work on affective polarization and intervention effects such as Mosleh2024-op. Its emphasis on deliberative ideals as implicit normative backdrop is relevant to studies of online deliberation and discourse quality like Esau2025-tf and Rossini2026-jn, while its concern with misinformation-based recommendations connects to Starbird2025-jj and Efstratiou2025-gs.

Podcast

A research-radio episode discusses this paper: Listen