16:50 - 18:30
P10-S256
Room: 0A.09
Chair/s:
Matthew Tyler
Discussant/s:
Alexandra Jabbour
Assessing Policy-Driven Shifts in Public Opinion Across Nations: The Challenges of Staggered Adoption, Measurement Error, and Limited Sample Size
P10-S256-2
Presented by: Piotr Koc
Piotr Koc 1, Marco Steenbergen 2
1 GESIS – Leibniz Institute for the Social Sciences
2 University of Zurich
This paper proposes a method to investigate how public opinion responds to policy changes. Existing approaches have often relied on two-way fixed effects or hierarchical models with the Mundlak device. However, these methods can yield biased estimates when the timing of treatment varies across units. In response, new difference-in-differences estimators have been introduced to address the problem of staggered adoption. While these new approaches are flexible, they have not tackled two additional challenges commonly encountered when studying policy effects on public opinion across nations: first, that public opinion is not directly observed and must be inferred from noisy measurements; second, that cohorts defined by the policy introduction year can be extremely small, sometimes consisting of a single country. To address these issues, we propose a Bayesian hierarchical model that explicitly accounts for known measurement error and partially mitigates small-sample concerns. We detail the model’s underlying assumptions and limitations, and then present a small simulation study demonstrating its performance. Finally, we apply our approach to the case of same-sex couples’ rights in Europe, illustrating how it can improve inference about the relationship between policy changes and shifts in public opinion.
Keywords: Bayesian, public opinion, measurement error, causal inference

Sponsors