Header

UZH-Logo

Maintenance Infos

Learning from polls


Leemann, Lucas; Stoetzer, Lukas; Traunmueller, Richard (2020). Learning from polls. In: EPSA Annual Conference, Virtual, 18 June 2020 - 19 June 2020, 1-21.

Abstract

Voters’ expectations of party strengths are a central part of many foundational political science theories that posit a strategic act by the voter. But how do voters develop these beliefs and how is this belief formation affected by polling reports? In this article, we present a dynamic Bayesian learning model that serves as a baseline for how beliefs are formed. We use survey experiments to estimate parameters of the dynamic learning process and analyze how and when belief formation deviates from theoretical model. We find that respondents update closely to new arriving poll results, they judge the polls to be two times more imprecise as the actual sample error and that this makes the induced differences in prior beliefs about a race vanish over time. We further apply the experiment to the study of partisan bias and the quality of the polls.

Abstract

Voters’ expectations of party strengths are a central part of many foundational political science theories that posit a strategic act by the voter. But how do voters develop these beliefs and how is this belief formation affected by polling reports? In this article, we present a dynamic Bayesian learning model that serves as a baseline for how beliefs are formed. We use survey experiments to estimate parameters of the dynamic learning process and analyze how and when belief formation deviates from theoretical model. We find that respondents update closely to new arriving poll results, they judge the polls to be two times more imprecise as the actual sample error and that this makes the induced differences in prior beliefs about a race vanish over time. We further apply the experiment to the study of partisan bias and the quality of the polls.

Statistics

Downloads

3 downloads since deposited on 14 Jan 2021
3 downloads since 12 months
Detailed statistics

Additional indexing

Item Type:Conference or Workshop Item (Paper), not_refereed, original work
Communities & Collections:06 Faculty of Arts > Institute of Political Science
Dewey Decimal Classification:320 Political science
Language:English
Event End Date:19 June 2020
Deposited On:14 Jan 2021 10:09
Last Modified:15 Jan 2021 04:38
Additional Information:10th Annual Conference, 18-19 June 2020
OA Status:Closed
Related URLs:https://www.epsanet.org/conference-2020/ (Organisation)

Download

Closed Access: Download allowed only for UZH members

Content: Submitted Version
Language: English
Filetype: PDF (Slides) - Registered users only
Size: 747kB
Content: Accepted Version
Language: English
Filetype: PDF - Registered users only
Size: 988kB