The increasing popularity of strongman rule in democratic societies underscores the need to explore how authoritarian regime preferences might spread socially. We assess the role of social influence on support for leaders with authoritarian inclinations through preregistered field and survey experiments in the Norwegian Armed Forces. The field experiment randomly assigned soldiers to different rooms during boot camp, so soldiers lived among peers with varying levels of openness to authoritarian rule. We found that many individuals adjusted their privately reported support for authoritarian rule to align more closely with their peers. Further survey-experimental evidence among soldiers and the general Norwegian population confirms that learning about others’ level of support for authoritarian rule changes both perceptions and own support. Our results suggest that support for authoritarian rule can have a social basis and could potentially spread through social contagion in established democracies.

https://onlinelibrary.wiley.com/doi/10.1111/ajps.70067

Public opinion research has long faced a trade-off between depth and scale: standardized surveys enable large-scale measurement but restrict respondents to researcher-defined categories, obscuring the diversity of unexpected considerations that underlie public sentiment. More conversational interviews provide richer insights through open-ended probing, but their reliance on trained human interviewers has kept them difficult to scale. This study introduces AI Conversational Interviewing as a method for collecting open-ended public opinion data at scale, pursuing three objectives: to demonstrate the analytical value of conversational text data for questions beyond the reach of closed-ended items; to assess the method’s practical viability through participants’ own evaluations; and to inform implementation by experimentally comparing voice-based, chat-based, and free-choice interview modes. We conducted a study combining an AI-led interview with a standardized survey on migration policy among 571 respondents recruited via Prolific and Payback Panel. The findings establish AI Conversational Interviewing as a viable and valuable addition to the social-science toolkit. The conversational transcripts surface considerations and reasoning that a comprehensive standardized battery does not capture such as markedly different mental models of migration among subgroups with similar attitudes levels. Among respondents who completed the interview, evaluations of the AI interview were at or above those of the standardized survey across modes, although completion itself varied by condition. By releasing open data and open-source pipeline materials, the study contributes to a growing literature on harnessing artificial intelligence to expand the methods of public opinion measurement.

https://arxiv.org/abs/2606.20064

Why do people profess support for democracy yet vote for politicians who undermine it? Alexander Wuttke, Junior Professor of Digitalization and Political Behavior at LMU, is tackling this question with a €1.17 million Emmy Noether grant from the DFG. His project, Predictably Paradoxical: Leveraging AI to Map the Democratic Mind, uses AI-supported interviews to explore how contradictory democratic attitudes arise, combining large-scale surveys with in-depth, human-like conversations across 13 countries.

Read the full article on the LMU website.

The project addresses a key challenge of contemporary democracies: democratic backsliding increasingly occurs through elections and the decisions of ordinary citizens. It therefore examines how democratic support can be strengthened through effective communication.

Adopting a multidimensional view of democratic support, the project develops and tests theory-driven interventions tailored to different attitudinal profiles. These interventions are evaluated using large-scale survey experiments, in-depth online conversations, and in-person group discussions.

The project aims to generate practical insights into how pro-democratic communication can strengthen democratic resilience at the citizen level.

Teaser: Countries are eager to use AI to automate some government processes, but this risks eroding citizens’ trust and feelings of democratic control – because AI mistakes can ruin their lives.

Link

Read the full interview [here].

Link to pre-print

Abstract

The rise of many-analysts studies has highlighted substantial variability in research outcomes when different teams independently analyze the same datasets and hypotheses. This paper makes two key contributions to this emerging field. First, we demonstrate how the many-analysts framework can be integrated into research methods education, by having students act as independent analysts addressing the same research question, followed by a meta-analysis of their results. Enriching the pedagogical toolkit, this didactic approach teaches concepts like researcher degrees of freedom and the importance epistemic humility. Second, this study applies the many-analysts framework to causal inference using observational data to identify outcome variability in this area of research. Specifically, we engage students in independent difference-in-differences (DiD) analyses focused on local governance. The study reveals a wide range of effect sizes and divergent conclusions from a typical DiD design, emphasizing the benefits of robustness checks involving multiple independent analysts in research and teaching.

Link to the Study

Addressing a Subtle Threat to Democracy

In many countries, democratic backsliding no longer happens through overt authoritarian takeovers, but rather through democratic elections and the choices of ordinary citizens. This subtle erosion of democracy highlights the need to better understand and reinforce democratic commitment among the public.

A Multi-Dimensional Approach to Democratic Support

Our project is based on the premise that support for democracy is not a binary attribute but a multi-dimensional construct. Citizens may strongly support some democratic principles while holding reservations about others. These fissures in democratic attitudes require different communicative responses tailored to individual profiles.

Evidence-Based Democratic Persuasion

The project draws on cutting-edge research in psychology and political communication to design and test theory-driven interventions. We will explore the effectiveness of these interventions through three complementary experimental approaches:

• Large-scale survey experiments

• Introspective one-on-one Zoom conversations

• In-person group discussions replicating everyday political dialogue

The goal is to provide robust, actionable insights on how communication can effectively strengthen democratic orientations and counteract anti-democratic narratives.

Societal Impact

Our findings are intended to support politicians, civil society organizations, educators, and journalists in developing and implementing communication strategies that bolster democratic resilience. By empirically testing what works, this project contributes to the broader effort to defend democracy from within.

Join Our Team

To support this endeavor, we will soon be recruiting a PhD student and a Postdoctoral Researcher, each for a duration of three years. The positions will focus on Political Psychology, Experimental Methods, or Computational Methods. Job advertisements will be published in the coming weeks.

Applications for the first cohort are now open!

Prospective students are warmly invited to join our upcoming information session, where we will introduce the program structure, content, and application process, and answer your questions.

For more information about the program and the info session, please visit our website:
https://www.sw.lmu.de/en/studiengaenge/master-computational-social-science/

The program is jointly coordinated by Prof. Dr. Mario Haim, Prof. Dr. Carsten Schwemmer, and Prof. Dr. Alexander Wuttke.

https://arxiv.org/abs/2410.01824

Traditional methods for eliciting people’s opinions face a trade-off between depth and scale: structured surveys enable large-scale data collection but limit respondents’ ability to voice their opinions in their own words, while conversational interviews provide deeper insights but are resource-intensive. This study explores the potential of replacing human interviewers with large language models (LLMs) to conduct scalable conversational interviews. Our goal is to assess the performance of AI Conversational Interviewing and to identify opportunities for improvement in a controlled environment. We conducted a small-scale, in-depth study with university students who were randomly assigned to a conversational interview by either AI or human interviewers, both employing identical questionnaires on political topics. Various quantitative and qualitative measures assessed interviewer adherence to guidelines, response quality, participant engagement, and overall interview efficacy. The findings indicate the viability of AI Conversational Interviewing in producing quality data comparable to traditional methods, with the added benefit of scalability. We publish our data and materials for re-use and present specific recommendations for effective implementation.

Link

This study investigates researcher variability in computational reproduction, an activity for which it is least expected. Eighty-five independent teams attempted numerical replication of results from an original study of policy preferences and immigration. Reproduction teams were randomly grouped into a ‘transparent group’ receiving original study and code or ‘opaque group’ receiving only a method and results description and no code. The transparent group mostly verified original results (95.7% same sign and p-value cutoff), while the opaque group had less success (89.3%). Second-decimal place exact numerical reproductions were less common (76.9 and 48.1%). Qualitative investigation of the workflows revealed many causes of error, including mistakes and procedural variations. When curating mistakes, we still find that only the transparent group was reliably successful. Our findings imply a need for transparency, but also more. Institutional checks and less subjective difficulty for researchers ‘doing reproduction’ would help, implying a need for better training. We also urge increased awareness of complexity in the research process and in ‘push button’ replications.

Deutsch

English

Alexander Wuttke, Matthias Aßenmacher, Christopher Klamm, Max M. Lang, Quirin Würschinger, Frauke Kreuter

Traditional methods for eliciting people’s opinions face a trade-off between depth and scale: structured surveys enable large-scale data collection but limit respondents’ ability to express unanticipated thoughts in their own words, while conversational interviews provide deeper insights but are resource-intensive. This study explores the potential of replacing human interviewers with large language models (LLMs) to conduct scalable conversational interviews. Our goal is to assess the performance of AI Conversational Interviewing and to identify opportunities for improvement in a controlled environment. We conducted a small-scale, in-depth study with university students who were randomly assigned to be interviewed by either AI or human interviewers, both employing identical questionnaires on political topics. Various quantitative and qualitative measures assessed interviewer adherence to guidelines, response quality, participant engagement, and overall interview efficacy. The findings indicate the viability of AI Conversational Interviewing in producing quality data comparable to traditional methods, with the added benefit of scalability. Based on our experiences, we present specific recommendations for effective implementation.

Arxiv-Link, NotebookLM Podcast

Link

Link

Link

Abstract

Ordinary citizens can serve as a critical defence against democratic backsliding. But beneath the surface, citizens’ commitment to democracy is sometimes fragile, with crises exacerbating existing anxieties. We introduce ‘democratic persuasion’ as an actionable intervention to foster the resilience of citizens’ commitment to liberal democracy. ‘Democratic persuasion’ seizes the opportunity of communicating with wavering democrats. ‘Democratic persuasion’ entails actively making the case for democracy and discussing democracy’s inherent trade-offs while engaging existing doubts and misperceptions. Amid the COVID-19 pandemic, which stirred frustrations with democracy and highlighted democratic trade-offs, we invited citizens via Facebook to participate in one of sixteen Zoom town halls to engage in discussions on pandemic politics with members of German state and federal parliaments. Each representative hosted two town halls, with random assignment to a condition of ‘democratic persuasion’ in one of the two town hall meetings. The field experiment yielded mixed results, demonstrating significant effects on some indicators of democratic commitment but not on others. This study contributes to the nascent body of research aimed at reinforcing the societal pillars of liberal democracies.

Replication material, Pre-registration, Pre-Print, Study.