Research
CDI organizes its work around four connected themes—the mechanisms through which democracies maintain, lose, or regain integrity. Each member runs an independent research stream, but the center’s structure enables integrated studies, larger-scale data collection, and computational projects no single lab could pursue alone.
Misinformation & Belief Formation
Why do people accept or reject political information—and what actually changes their minds?
People form beliefs in complex, fast-moving media environments where accuracy is hard to verify. This theme uses experiments, natural language processing, and computational modeling to trace how misperceptions take hold and how susceptibility develops. It also builds and tests interventions—prebunking, corrections, and belief-revision strategies—that hold up in the field.
Representative work: Gollwitzer et al., Intellectual humility as a tool to combat false beliefs (British Journal of Social Psychology, 2024) · Gollwitzer et al., Partisan differences in physical distancing linked to health outcomes (Nature Human Behaviour, 2020). See Publications → Misinformation.
Polarization, Extremism & Radicalization
What drives division to escalate into hostility, extremism, and support for violence?
Identity, perceived threat, and grievance can turn ordinary disagreement into entrenched hostility. Drawing on large-scale surveys, experiments, and cross-cultural designs, this theme maps the psychology of dehumanization, relative deprivation, and radicalization—including a 2026 framework on radicalization in the age of AI. The aim is to identify the points where escalation can be interrupted.
Representative work: Kunst et al., Intelligent Systems, Vulnerable Minds: A Framework for Radicalization to Violence in the Age of AI (Personality and Social Psychology Review, 2026) · Kunst et al., Understanding Violent Extremism in the 21st Century (Current Opinion in Psychology, 2020). See Publications → Polarization.
Trust & the Stability of Institutions
Why do some democratic institutions hold under pressure while others give way?
Institutional trust underpins everything from turnout to compliance, and its erosion carries real economic and political cost. This theme integrates political economy, comparative politics, and organizational ethics to study how institutions respond to disinformation, extremist mobilization, and shifting norms. It connects micro-level behavior to macro-level outcomes—linking why individuals (dis)trust to whether institutions endure.
Representative work: Geys, Explaining voter turnout: a review of aggregate-level research (Electoral Studies) · Sitter, Democratic Backsliding in the European Union (Oxford Research Encyclopedia of Politics) · van Gils et al., Respectful leadership (Human Relations). See Publications → Trust.
AI & the Information Environment
How is AI reshaping political behavior, and how can we measure and counter its misuse?
Generative AI, synthetic media, and coordinated influence are changing how information is produced and spread—including “cyborg propaganda,” AI-written political content pushed through networks of real accounts. This theme builds AI-assisted tools to measure political language at scale and to detect misinformation and manipulation. It pairs technical detection with governance and responsible-AI research.
Representative work: Langguth et al., COCO: an annotated dataset of COVID-19 conspiracy theories (Journal of Computational Social Science, 2023) · Arnulf et al., Predicting Survey Responses (PLoS ONE) · Buhmann et al., Towards a deliberative framework for responsible innovation in AI (Technology in Society). See Publications → AI.
How we work
CDI’s differentiator is method. Where a typical policy shop summarizes, the center measures—with the tools its members build. This is what sets its research apart.
- Computational text analysis & LLM-based measurement Turning large, messy text—speeches, parliamentary records, social posts, media coverage—into structured signal, and building classifiers that flag false or manipulative content at scale.
- Experiments & multi-treatment intervention tests Rigorously testing what changes beliefs and behavior, including prebunking and correction strategies.
- Field studies & large-scale data Register-verified outcomes, real-world interventions, and cross-cultural samples.
- Institutional & comparative analysis Assessing how institutions respond to information threats and democratic backsliding.
- Network & detection methods Tracing coordinated influence and the spread of misinformation across platforms.
Publications
A curated, theme-grouped list lives on the Selected Publications page. Members’ full lists are on their Google Scholar profiles, linked on People.