実証政治学ワークショップのご案内（Workshop Announcement: April 21, 17:00-）

実証政治学ワークショップのご案内  (Workshop Announcement: April 30, 13:10-)

カターニア大学のCarammia Marcello准教授と、ハーバード大学IQSSのStefano Iacus氏をお迎えし、ワークショップを開催します。
どなたでもご参加いただけます。事前登録は不要です。ぜひご参加下さい。
We are very pleased to host Professor Carammia Marcello (University of Catania), and Director Stefano Iacus (IQSS, Harvard University) for a special talk. Pre-registration is not required. We look forward to your participation.

日時：2026年4月21日（火）17:00 pm – 18:40 pm
Date and Time: Tuesday, April 21, 2026, 17:00 pm – 18:40 pm

場所：3号館406
Venue：Room 406, Building 3, Waseda Campus, Waseda University

言語Language：英語English

<Carammia Marcello> 

タイトル Title： Policy Attention and Democratic Representation: Evidence from 30 Years of European Parliamentary Questions

要旨Abstract： Marcello Carammia (University of Catania), Federico Russo (University of Salento)
How well do elected representatives attend to the issues that matter to citizens? The question of policy responsiveness — whether legislators’ agendas reflect public priorities and real-world conditions — is central to theories of democratic representation, yet comparative empirical evidence remains limited by data availability, short time horizons, and institutional scope.
This paper advances the study of policy attention and representation by exploiting a uniquely advantageous setting: the European Parliament, where ~3,000 MEPs from 29 countries and nine party families have filed more than 180,000 written questions across six legislative terms (1994–2024). This institutional context offers exceptional variation in national backgrounds, partisan affiliations, and exposure to heterogeneous economic and social conditions — all within a single legislative body operating under common rules.
Using the Comparative Agendas Project coding framework applied via fine-tuned large language models, we first describe how policy attention is distributed and how it shifts over time and across crises. We then model the extent to which MEPs’ issue agendas respond to real-world problem indicators (unemployment, migration, economic performance) and to citizens’ stated priorities as captured by Eurobarometer “Most Important Problem” data. The analysis contributes to the broader literature on policy responsiveness and agenda-setting by providing large-N, longitudinal evidence on the conditions under which elected representatives’ attention tracks — or fails to track — the concerns of the public they serve.

<Stefano Iacus>

タイトル Title： Dynamic Attention (DynAttn): Interpretable High-Dimensional Spatio-Temporal Forecasting (with Application to Conflict Fatalities)

要旨Abstract： Dynamic Attention (DynAttn): Interpretable High-Dimensional Spatio-Temporal Forecasting (with Application to Conflict Fatalities)
 Forecasting conflict-related fatalities remains a central challenge in political science and policy analysis due to the sparse, bursty, and highly non-stationary nature of violence data. We introduce DynAttn, an interpretable dynamic-attention forecasting framework for high-dimensional spatio-temporal count processes. DynAttn combines rolling-window estimation, shared elastic-net feature gating, a compact weight-tied self-attention encoder, and a zero-inflated negative binomial (ZINB) likelihood. This architecture produces calibrated multi-horizon forecasts of expected casualties and exceedance probabilities, while retaining transparent diagnostics through feature gates, ablation analysis, and elasticity measures.
We evaluate DynAttn using global country-level and high-resolution PRIO-grid-level conflict data from the VIEWS forecasting system, benchmarking it against established statistical and machine-learning approaches, including DynENet, LSTM, Prophet, PatchTST, and the official VIEWS baseline. Across forecast horizons from one to twelve months, DynAttn consistently achieves substantially higher predictive accuracy, with particularly large gains in sparse grid-level settings where competing models often become unstable or degrade sharply.
Beyond predictive performance, DynAttn enables structured interpretation of regional conflict dynamics. In our application, cross-regional analyses show that short-run conflict persistence and spatial diffusion form the core predictive backbone, while climate stress acts either as a conditional amplifier or a primary driver depending on the conflict theater.

共催：
早稲田大学現代政治経済研究所　世論・メディアデータ研究部会　
Public Opinion and Media Data Research Group, WINPEC, Waseda University

Contact:：
日野愛郎（早稲田大学） Airo Hino (Waseda University)  [email protected]