14:00-14:15
Welcome by Lisa Bruttel
14:15-15:45
Session 1
Joel Lamb (University of Exeter) “Strategy of the Commons: Causal Evidence from a Lottery in Parliament”
How do politicians secure top government positions? We argue that politicians make strategic decisions about the legislation they propose to advance their careers. We introduce a model of political capital that yields predictions about the legislation proposed by politicians. We exploit a natural experiment in the UK House of Commons, where politicians enter a lottery to win the opportunity to introduce legislation of their choosing. First, we verify that winning this lottery improves career outcomes, leading to a 71% increase in ministerial appointments five years after treatment. Next, we use Natural Language Processing techniques to analyse the content of the bills presented by ballot winners. We provide evidence that politicians who strategically use this opportunity to push party objectives see a larger boost to their careers.
Giacomo Manferdini (Bocconi University) “Blameocracy: Causal Attribution in Political Communication”
We propose a supervised method to detect causal attribution in political texts, distinguishing between expressions of merit and blame. Analyzing four million tweets shared by U.S. Congress members from 2012 to 2023, we document a pronounced shift toward causal attribution following the 2016 presidential election. The shift reflects changes in rhetorical strategy rather than compositional variation in the actors or topics of the political debate. Within causal communication, a trade-off emerges between positive and negative tone, with power status as the key determinant: government emphasizes merit, while opposition casts blame. This pattern distinguishes causal from purely affective communication. Additionally, we find that blame is associated with lower trust in politicians, perceived government effectiveness, and spreads more virally than merit.
15:45-16:30
Coffee break
16:30-18:00
Session 2
Gonzalo Arrieta (University of Zurich) “Procedural Decision-Making In The Face Of Complexity”
A large body of work documents that complexity affects individuals’ choices, but the literature has remained mostly agnostic about why. We provide direct evidence that individuals use different choice processes for complex and simple decisions. We hypothesize that individuals resort to “procedures”—cognitively simpler choice processes that we characterize as being easier to describe to another person—as the complexity of the decision environment increases. We test our hypothesis using two experiments, one with choices over lotteries and one with choices over charities. We exogenously vary the complexity of the decision environment and measure the describability of choice processes by how well another individual can replicate the decision-maker’s choices given the decision-maker’s description of how they chose. We find strong support for our hypothesis: Both of our experiments show that individuals’ choice processes are more describable in complex choice environments, which we interpret as evidence that decision-making becomes more procedural as complexity increases. We show that procedural decision-makers choose more consistently and exhibit fewer dominance violations, though we remain agnostic about the causal effect of procedures on decision quality. Additional secondary evidence suggests that procedural decision-making is a choice simplification that reduces the cognitive costs of decision-making.
Caterina Giannetti (University of Pisa) “Teaming up with artificial players”
Using a real-life escape room scenario, we investigate how different levels of embodiment in artificial agents influence team performance and conversational dynamics in non-routine analytical tasks. Teams composed of either three humans or two humans and an artificial agent (a Box, an Avatar, and a hyper-realistic Humanoid) worked together to escape the room within a time limit. Our findings reveal that while human-only teams tend to complete all tasks more frequently, they also tend to be slower and make more errors. Additionally, we observe a non-linear relationship between the degree of agent embodiment and team performance, with a significant effect on conversational dynamics. Teams with agents exhibiting higher levels of embodiment display conversational patterns more similar to those occurring among humans. These results highlight the complex role that embodied AI plays in human-agent interactions, offering new insights into how artificial agents can be designed to support team collaboration in problem-solving environments.
9:00-10:00
Keynote lecture by Gloria Gennaro (University College London) “Emotion and Reason in Political Language”
Introduction by Maximilian Andres
This talk covers two related projects that investigate the use and drivers of emotional rhetoric in U.S. congressional discourse and its evolution over time. Using computational linguistics, we develop a validated, text-based scale of emotionality and apply it to over six million congressional speeches from 1858 to 2014. We document sharp increases in emotional appeals beginning in the late 1970s, particularly among Democrats, women, minorities, ideologically extreme legislators, and members of the opposition. Next, we examine how the introduction of televised floor debates via C-SPAN may have contributed to the rise of emotional rhetoric in the last few decades. A difference-in-differences analysis reveals that the 1979 launch of C-SPAN significantly increased emotional rhetoric in the House relative to the Senate. Using channel positioning as an instrument for viewership, we show that representatives from districts with higher C-SPAN exposure use more emotional appeals, especially in competitive districts. However, contrary to standard transparency-based accountability models, C-SPAN has no positive effect on legislative effort, and may even reduce constituency orientation. In contrast, local news coverage increases legislative effort without affecting rhetorical style. Finally, we find that C-SPAN exposure raises incumbents' vote shares, particularly for those who employ more emotional language. These findings underscore the growing role of emotion in legislative discourse and reveal how transparency, mediated or direct, shapes political behavior through audience incentives.
10:00-10:30
Coffee break
10:30-12:00
Session 3
Lara Berger (University of Cologne) “How digital media markets amplify news sentiment”
This paper investigates whether digitization, by increasing incentives to capture reader attention, leads to more emotional news headlines. Comparing online and offline editions of major newspapers, I find that online headlines are significantly more emotional—particularly negative—than their print counterparts. Two experiments with professional journalists further suggest that these differences arise from stronger economic incentives: when journalists are compensated based on click or subscription rates, they systematically choose more emotional headlines. A connected reader experiment reveals that such sensational framing increases reading times but reduces factual retention, highlighting potential social costs of an intensified emotional slant in digital news environments.
Prashant Garg (Imperial College London) ”Political Expression of Academics on Social Media”
Academics have traditionally played a vital role in both the generation and dissemination of knowledge, ideas and narratives. Social media, relative to traditional media, provides for new and more direct ways of science communication. Yet, since not all academics may engage with social media, the sample that does so may have an outsize influence on shaping public perceptions of academia more broadly through at least two channels: the set topics they engage with and through the particular style and tone of communication. This paper describes patterns in academics' expression online found in a newly constructed global dataset covering over 100,000 scholars linking their social media content to academic record. We document large and systematic variation in politically salient academic expression concerning climate action, cultural, and economic concepts. We show that these appear to often diverge from general public opinion in both topic focus and style.
13:00-15:15
Session 4
Magdalena Wasilewska (University of Amsterdam) ”Explanations of Inequality”
Economic inequality has become an important element of political debates, capturing significant attention in media coverage, public discourse, as well as academic research. In a survey experiment on a representative sample of the Dutch population (N = 4501), we examine how individuals explain economic inequality. We employ responses from an open-ended survey question about sources of inequality and link them on an individual level to administrative data on respondents’ income and wealth. We find that respondents predominantly attribute inequality to structural factors rather than personal responsibility or luck. Explanations are somewhat insensitive to information about inequality in the Netherlands, or about one’s standing in the income or wealth distributions. We observe significant differences in the content of the explanations based on respondents’ characteristics, such as wealth or political affiliation, and inequality-related attitudes. In a subsequent data collection, we aim to estimate the causal impact of explanations of inequality on redistributive preferences.
J. Jobu Babin (DePaul University) “Text Analysis of Subjects’ Perceptions of One-shot Trust Games”
The Trust Game (TG) is a widely used experimental paradigm to study trust and prosocial behavior, yet its interpretation remains debated. This study leverages natural language processing and language AI to analyze how subjects conceptualize TG interactions and predict behavior. Using a novel dataset and an experimental setup across three binary variants, we examine how participants explicitly frame their perception of the interaction and subsequent decisions. BERTopic modeling and ensemble machine learning models reveal that themes involving "trust" and "working together" strongly predict passing behavior, while risk-related and game-theoretical concepts play a lesser role. These findings both validate and challenge conventional assumptions, highlighting the importance of social context in economic decision-making.
Can Celebi (VCEE, University of Vienna) “Using Large Language Models for Text Classification in the Social Sciences”
This study examines the use of large language models (LLMs) for text classification. We investigate whether original instructions can be effectively repurposed as prompts with minimal changes to achieve classification results comparable to human-coded benchmarks. Additionally, we study the impact of two prompting techniques – varying the number of classified examples (n-shot) and requiring a justification explanation (zero-shot Chain-of-Thought) – on classification performance. Using GPT-3.5 and GPT-4, we further examine the extent to which larger model size improves classification accuracy. To assess these factors, we classify text from four economic experiments, covering tasks with varying complexity and prevalence in pre-training data, providing insights into how task characteristics influence classification performance. Our findings offer guidance for integrating LLM-based text classification into social science research.
15:15-15:30
Goodbye by Vasilisa Werner