Am Donnerstag, 09.01.2025 ist Saskia Bartholomäus (Gesis Institut) in meinem Kolloquium zu Gast. Sie hat im September 2022 ihr Studium der empirischen Demokratieforschung abgeschlossen und ist seitdem wissenschaftliche Mitarbeiterin bei Gesis, wo sie zu Themen aus der Umfrageforschung promoviert.
Von 12 bis 14 Uhr spricht sie in Raum 02-601 (Georg-Forster-Gebäude) zum "Thema Non-response Bias" in der politikwissenschaftlichen Umfrageforschung. Alle Interessierten sind herzlich eingeladen!
Abstract:
The under-representation of politically disengaged persons in political science surveys limits analyses and strains conclusions that can be drawn from the data. Controlling for non-response biases through weighting procedures does not overcome the limited number of politically disengaged respondents and, thus, the low variation in political engagement variables. Instead, Adaptive Survey Designs that vary the survey protocol across subgroups of respondents to optimize their response rates offer the potential to improve political science surveys non-response bias. Political scientists could leverage respondents' interest in different topics to improve their survey participation. In practical terms, politically disengaged respondents would receive a questionnaire combining political and non-political modules, while politically engaged respondents would receive a purely political questionnaire. To avoid biases in the varying question module due to selective assignment, researchers could assign only a certain percentage of respondents with low political engagement to tailored questionnaires and use missing data strategies based on data from the main question module to correct biased estimates.
In this presentation, I will answer my dissertation's overarching research question: How to reduce political science surveys’ non-response bias without reducing data quality by accounting for respondents’ political interest? The dissertation consists of four separate studies that focus on different aspects that need to be investigated in order to answer my research question:
- “Increasing Bias due to Negative Survey Experiences: Causes of Politically Disengaged Respondents’ Attrition in Political Science Panel Surveys”
- “Varying the Content of Political Science Surveys to Improve the Survey Experience Among Politically Disengaged Respondents”
- “Comparing Different Approaches to Reduce Attrition of Political Disengaged Respondents in Political Science Surveys. Potentials and Risks”
- “Reducing Political Science Surveys’ Attrition Bias Without Limiting Substantive Research: Combining Adaptive Survey Design and Missing Data Strategies”
Most analyses are based on data simulations and a survey experiment conducted in the probability-based GESIS Panel. My findings suggest that Adaptive Survey Designs that consider respondents' political interests by including non-political content are likely to reduce the non-response bias of political science surveys. However, to maintain the data quality for each survey question module, only a certain percentage of respondents should be allocated to different question modules according to their interests, allowing missing data strategies to correct for induced selection biases. This approach may be promising for addressing non-response bias in cross-sectional and longitudinal, probability- and nonprobability-based political science surveys, allowing closer analyses of the causes and consequences of political disengagement.
