🕑 Join us *today* at 13:45 CEST for the talk 'Harnessing GPT for Enhanced Academic Writing: Evidence from a Field Experiment with Early-Career Researchers in the Social Sciences' by
Marc Ratkovic (University of Mannheim)
⬇️ Zoom link and details below
![2026-09-16 | Input talk | Marc Ratkovic (University of Mannheim)
Harnessing GPT for Enhanced Academic Writing: Evidence from a Field Experiment with Early-Career Researchers in the Social Sciences
more
Hybrid event [A5, 6, Room A231 + Zoom]
September 16, 2026, 13:45-15:15
Abstract
Large Language Models (LLMs) bring both risk and opportunity to scholarly communication, yet their effects on expert-level research remain untested. We conducted the first randomized controlled trial of GPT-4 in expert academic writing, implemented as a three-day hackathon (n=22 early-career social scientists) in collaboration with Springer Nature. Participants were randomized to unrestricted GPT access or a no-AI control. Manuscripts were evaluated pre-and post-intervention by expert faculty and postdoctoral researchers on six dimensions: clarity, coherence, originality, methodological rigor, depth of analysis, and literature relevance. AI assistance produced significant improvements in clarity (Δ=+0.49, p=0.009) and coherence (Δ=+0.43, p=0.036) without affecting other dimensions. Principal-components analysis confirmed AI's selective enhancement of organizational features. SakanaAI, a GPT-based reviewer, aligned broadly with human assessments. A linguistic analysis of participants' reflective journals revealed increased process-oriented language and reduced temporal references, suggesting a reallocation of cognitive resources. Although concerns about overreliance and academic integrity persist, our findings imply that delegating routine structuring tasks to AI can free scholars to focus on generating novel insights.
Presenter(s)
Marc Ratkovic is Professor of Political Science and Social Data Science at the University of Mannheim and Co-Founder and Co-Director of the Global Responsible Tech Lab. Previously at Princeton University, his research spans political methodology, causal inference, and machine learning and has appeared in leading journals in political science and statistics.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:oizrkjzyr6vekl76tjctghcl/bafkreifpssyfllrstbpry2zsxhsgb3fg4y5cki5vm2hjyikf6mi4paab2a)
![Survey Recruitment on LinkedIn: A Step-by-Step Guide to Targeted Outreach
March 25, 2026 31 min read tutorials [Zaza Zindel Lisa de Vries]
LinkedIn’s professional user base and precise targeting capabilities present social science researchers with a valuable opportunity to recruit survey participants from highly specific populations, particularly those relevant to research on employment and industry dynamics. While platforms such as Facebook and Instagram are commonly used for recruitment of survey participants, LinkedIn’s potential in survey participant recruitment remains underutilized. Its professional focus, career-centric context, and only minimal off-topic content make it ideal for reaching the potential labor force population. This is particularly relevant for studies requiring insights from specific industries, occupations, or education levels. In this Methods Bites Tutorial, Dr. Zaza Zindel (German Centre for Integration and Migration Research (DeZIM)) and Dr. Lisa de Vries (FernUniversität in Hagen) provide a step-by-step guide on how to use LinkedIn ads to recruit participants for survey research.
Continue reading
Social media Survey data Data collection](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:oizrkjzyr6vekl76tjctghcl/bafkreictz42gbz2dbeds5r26mqj62rmnhpdpctp3t2rrwd4gszeehmv4x4)
![2026-04-22 | Input talk | Leah von der Heyde (GESIS)
AIn't Nothing But a Survey? Using Large Language Models for Coding Open-Ended Survey Responses
Hybrid event [A5, 6, Room A231 + Zoom]
April 22, 2026, 13:45-15:15
Abstract
Large language models (LLMs) have the potential to make survey research more efficient, including the classification of open-ended survey responses. However, as most existing research on this topic has focused on English-language text or single LLMs, it is unclear whether their applicability generalizes and how the quality of classifications compares to established methods. In this talk, I will demonstrate how LLMs can be used for coding open-ended responses using different access options and prompting and fine-tuning techniques. I will present a study testing these approaches on a dataset of German open-ended survey responses, comparing several LLMs to human coders and other automated methods. Finally, I will discuss the implications of the study findings for practitioners, including the many trade-offs researchers need to consider.
Presenter(s)
Leah von der Heyde is a computational social scientist and survey methodologist. Her research focuses on the potential and pitfalls of new data sources, such as large language models, for improving the measurement and representation of public opinion. Substantively, she is particularly interested in political attitudes and voting behavior. Leah received her PhD in Social Data Science and Research Methodology from the University of Mannheim. She has a background in political science from LMU Munich, the University of Mannheim, and Georgetown University. Previously, Leah was part of the Social Data Science and AI Lab at LMU Munich, worked for the European Social Survey at GESIS - Leibniz Institute for the Social Sciences, the European Parliamentary Research Service, and several market and public opinion research institutes in Germany and Sweden. At GESIS, she is part of KODAQS, researching and educating…](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:oizrkjzyr6vekl76tjctghcl/bafkreigb4xgjzges3yeoup7w3z3goxg6fgei4qiystwxbq4zkvjughks6m)
![2026-03-11 | Input talk | Andreas Küpfer (TU Darmstadt & University of Birmingham)
Politics in Action: Studying Multimodal Data from Local Meetings to National Parliaments
Hybrid event [A5, 6, Room A231 + Zoom]
März 11, 2026, 13:45-15:15
Abstract
The analysis of multimodal political communication enables political scientists to study how politicians and parties behave in previously untapped ways. While large volumes of audiovisual data are increasingly available, these data are typically unstructured and vary substantially in quality, requiring robust and carefully designed processing pipelines. This talk guides through the challenges and opportunities of analysing political video data across institutional contexts. Analysing videos from local political meetings, including US school boards and state supreme courts, allows the generation of features such as transcriptions, gendered speech shares, and agenda items. However, the often semi-professional production quality of such recordings limits the systematic use of visual information. By contrast, legislative debate recordings are typically standardised and post-edited, enabling the leverage of visual features such as eye contact seeking to study the emergence of dominant behaviour during parliamentary speech. Scaling multimodal processing and feature extraction workflows across heterogeneous institutional contexts is, therefore, crucial. It enables the systematic study of political communication across levels of government and over longer time periods.
Presenter
Andreas Küpfer is a Postdoctoral Researcher in the DFG-funded project "Analysing Legislative Politics with Aligned Text, Audio and Image Data" at the Institute for Political Science of the Technical University of Darmstadt, holding a Ph.D. in Political Science from the University of Mannheim. He works at the intersection of Data Science and Political Science. His research focuses on (multimodal) political communication, combining computational with tradi…](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:oizrkjzyr6vekl76tjctghcl/bafkreiawvyqdytqoc4essm7llpzps33qsxpv6spg6i4fi5fk7qy33p63ye)
![2026-02-25 | Input talk | Klara Müller (University of Mannheim)
When Events Change Samples: Disentangling Causal Effects from Compositional Bias in Quasi-Experimental Designs
Hybrid event [A5, 6, Room A231 + Zoom]
Februar 25, 2026, 13:45-15:15
Abstract
From political psychology, we know that salient political events can shape political attitudes and behavior. In studying such event effects, quasi-experimental designs, such as the unexpected event during survey design (UESD), have become increasingly popular. To allow for causal identification, these designs rest on quite strict assumptions, one of which is the random assignment of respondents to the pre- and post-event sample and the comparability of these samples. From survey methodology, however, we also know that external events can affect who responds to surveys. If event-triggered shifts in survey participation and sample composition are related to outcome variables of interest, causal estimates of an event’s effect may be biased. This problem is particularly challenging when compositional shifts involve unobserved or even unobservable factors. In this talk, I provide intuition for how such compositional bias can threaten causal conclusions in quasi-experimental settings. I present a framework to disentangle an event’s “true” causal effect from compositional bias, with a particular focus on the UESD approach. The framework outlines practical strategies to adjust for observable imbalances and extends sensitivity analyses to assess how strong unobserved confounders would have to be to change or undermine substantive causal conclusions. I illustrate the approach using the rally-around-the-flag effect following the 2015 Charlie Hebdo attacks in France, as well as replications of published UESD studies on terrorist events and rally-style outcomes. By addressing both observable and unobservable sources of bias, this framework enhances causal inference and strengthens the credibility of public opinion research in d…](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:oizrkjzyr6vekl76tjctghcl/bafkreic2cgit5plcrxatdivgtpsgu3lcasarcxqw33t5no6a7fzcp2smyi)
![Brave New Data Access World: What the Digital Services Act (DSA) Means for Researcher Access to Digital Platforms
Hybrid event [A5, 6, Room A231 + Zoom]
Dezember 03, 2025, 13:45-15:15
Abstract
The European Union’s Digital Services Act (DSA) marks a turning point for the study of digital platforms by establishing a legal framework for researcher access to platform data, including from major social media companies. Article 40 of the DSA outlines how researchers can request such access, opening unprecedented possibilities for empirical investigations into the digital public sphere. Yet, the practical implementation of this right to data access raises important challenges. Questions of eligibility, the scope of accessible data, and the ethical and technical standards required for secure processing are central to understanding what this new regime will mean in practice. This talk provides an overview of the DSA’s data-access provisions, explains their relevance for the research community, and discusses what opportunities—and limitations—emerge for studying systemic risks, algorithmic effects, and platform governance. It reflects on early experiences with Article 40 implementation and considers what it will take to build a sustainable infrastructure for independent, policy-relevant research on digital platforms in Europe.
Presenter
Jakob Ohme leads the interdisciplinary "Digital News Dynamics" research group at the Weizenbaum Institute in Berlin, exploring the impact and dissemination of professional journalism on digital platforms versus other sources like influencers or AI. His work emphasizes the changes digital and mobile communications bring to news consumption and political engagement, particularly across different generations. Jakob Ohme is dedicated to advancing digital methodologies in political communication and journalism research, notably through the innovative use of digital trace data. He's also a Co-Principal Investigator in the #DSA40 Collaboratory, foc…](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:oizrkjzyr6vekl76tjctghcl/bafkreiel7po2hq6ec4nl66aupw2wper4yrmr4dywhmbeihjiki2mapsowe)
![What Shapes Voter Perceptions of Election Polls? The Impact of Partisanship and Visual Design
Hybrid event [A5, 6, Room A231 + Zoom]
November 05, 2025, 13:45-15:15
Abstract
Election polls are a central feature of modern electoral campaigns, providing voters with information that can guide their decisions. Yet, voters are often confronted with many polls showing volatile or conflicting outcomes, and media outlets present these polls in very different ways. This raises two key questions: (1) How do voters judge the trustworthiness of a poll’s results? and (2) How does the way a poll is presented affect these perceptions? I present experimental evidence from two large-scale studies in Germany. Together, they show that both the content of a poll and the way it is presented influence voter perceptions, offering new insights into how individuals navigate modern campaigns environments.
Presenter(s)
Elias Koch is a Doctoral Researcher at the Research Training Group DYNAMICS (HU Berlin/Hertie School) and a Research Associate at the Hertie School in Berlin. His research explores party competition and political behaviour in established democracies, with a particular focus on shifting public opinion landscapes.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:oizrkjzyr6vekl76tjctghcl/bafkreiea2wek2lsf5fbeqvhbk7bxgtzh6nvd6oodcuvj42nsza6uztkrma)
![Large Language Models for Social Research: Potentials and Challenges
Hybrid event [A5, 6, Room A231 + Zoom]
Oktober 22, 2025, 13:45-15:15
Abstract
Large Language Models (LLMs) have the potential to revolutionize the social sciences—for example, by accelerating content analysis or enabling realistic social simulations. In this workshop, I will discuss how LLMs can be applied and audited for social science applications, including the generation of synthetic survey responses and content analysis. I will also address how biases in LLMs can hinder these applications and explore ways to better surface and understand these biases. Finally, I will present a hands-on use case demonstrating how LLMs can be guided using demographic personas for both content analysis and simulated surveys.
Presenter(s)
Indira Sen is a Junior Faculty member at the University of Mannheim’s Business School in the Chair of Data Science for the Social and Economic Sciences. Her work lies at the intersection of NLP and Computational Social Science, specifically in developing and evaluating representative and equitable language technology, including Large Language Models.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:oizrkjzyr6vekl76tjctghcl/bafkreigj6rty2hy4hgawqzq34a24si45dxzutezuphif3hprnl5amxldoy)
![Roundtable Discussion: New Developments in Large-Scale Survey Data in Germany
Hybrid event [A5, 6, Room A231 + Zoom]
September 24, 2025, 13:45-15:15
Abstract
The roundtable brings together researchers from leading institutes and survey programs in Germany including the German Longitudinal Environmental Study (GLEN), the Family Research and Demographic Panel (FreDA), the Research Data Center of the Institute for Employment Research (IAB), and the Survey of Health, Aging and Retirement in Europe (SHARE). The discussion will focus on current innovations, challenges, and opportunities in large-scale survey data from the data producers’ and users’ perspective.
Presenter(s)
Claudia Schmiedeberg is a postdoctoral researcher at the Department of Sociology at LMU Munich. Her research focuses on survey methodology, environmental topics, and couple relationships.
Pablo Christmann is the project coordinator of FReDA – The German Family Demography Panel Study and a postdoctoral researcher at GESIS. His main research interests include political attitudes as well as survey methodology and methods.
Stefanie Wolter is a senior researcher at the Research Data Center of the Federal Employment Agency. She is project head of the Linked Personnel Panel, and responsible for linking enterprise and establishment data. Her research focuses on flexible work and within-firm inequality.
Michael Bergmann is a survey methodologist with a doctorate in social sciences from the University of Mannheim. As part of a joint appointment by htw saar and SBI, he works as head of the Survey Methodology department for the Survey of Health, Ageing and Retirement in Europe (SHARE) and as professor of survey methodology at the Faculty of Social Sciences. His research interests include methods for improving the quality of survey data, the investigation of the effects of different survey modes on data quality in panel studies, and the analysis of interviewer behavior. In terms of content, he is primarily …](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:oizrkjzyr6vekl76tjctghcl/bafkreigmoyska5t5x3cuyoxezazksxqcinxg7dgeoitepbcw4hftfylao4)
![Assessing the Reproducibility of Observational Social Research
Hybrid event [A5, 6, Room A231 + Zoom]
September 10, 2025, 13:45-15:15
Abstract
Reproducibility – i.e., the extent to which results are consistent when re-running the same code on the same data – is a minimum requirement for credible empirical research. However, as prior audits have tended to look at selective samples, only little is known so far about its extent in the observational social sciences. In this input talk, I present insights from a large-scale reproducibility assessment of published social science papers that use data from the European Social Survey (ESS). Starting from an initial pool of 1,206 articles, we obtained research code for 385 papers and conducted a standardized reproducibility assessment on a random sample of 100. I will present insights into our method and our results, highlight common hurdles and pitfalls we encountered, and conclude by discussing low-cost measures that could help improve reproducibility at scale.
Presenter(s)
Laura Schächtele is a PhD candidate in Sociology at the LMU Munich and part of the DFG-funded meta-scientific research program META-REP. She holds a master’s degree in Sociology from the LMU Munich and has been a visiting researcher at the Meta-Research Center at Tilburg University. In her dissertation, she investigates the links between academic incentive structures, researchers’ strategic behavior, and scientific transparency. Her broader research interests include the sociology of science, social inequality, and quantitative methods of empirical social research](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:oizrkjzyr6vekl76tjctghcl/bafkreie66sio5k5bkn7g7lhjn3dafo3eam5supiamg7jprotikx2qucgae)
![2025-03-05 | Input talk | Nicole Schwitter (University of Mannheim)
Digital Data, Contextual Complexities, and Lessons for Social Scientists
more
Hybrid event [A5, 6, Room A231 + Zoom]
March 05, 2025, 13:45-15:15
Abstract
With increasing digitalisation, data is everywhere - (seemingly) waiting to be harvested, by social scientists, from historic archives and search trends to Facebook friendships and open-source collaborations. Yet beneath this public surface lies the usual complexity of data generation, raising important questions about whether the observable actions represent meaningful concepts. This talk will serve as a cautionary reminder for social scientist to scrutinise their data, investigate and describe the data-generating context and community, and to (also) apply formal theory when measuring constructs. I will be drawing on several well-published examples and highlight their shortcomings to underscore these points and to suggest how we can – and need to – do better.
Presenter(s)
Nicole Schwitter is a postdoctoral researcher at the MZES and holds a PhD in Sociology from the University of Warwick, where she studied offline interactions in online communities using Wikipedia as a case study. Her research explores (inter)group relations and social dynamics, with a strong interest in applying novel computational methods to social science questions.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:oizrkjzyr6vekl76tjctghcl/bafkreia23ea3atn43u7wbnxxt6iep4zlolpapavu3hogio7s6zqanitxde)
