Prediction-based Adaptive Designs for Panel Surveys

• Applicants: Professor Dr. Tobias Gummer; Dr. Christoph Kern; Dr. Bernd Weiß
• Subject Area: Empirical Social Research
• Term: since 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 495763090

Despite its promising potential to reduce attrition and biases, the use of adaptive survey designs in panel studies is lacking in both areas that are needed for its functioning: (1) In predicting nonresponse and thus creating appropriate strata as well as (2) in the treatments that are administered in practice. This project will pair the implementation and testing of innovative prediction methodology from the field of machine learning with innovative treatments that can be assigned to likely nonrespondents. Prediction models will be trained and evaluated in a longitudinal framework that is tailored to identifying panelists at risk of nonparticipation in a given (new) panel wave. The predicted risk scores of the most accurate model allow us to test the effectiveness of different treatments. Specifically, this project will investigate the usage of innovative treatments in adaptive survey designs that aim to increase survey enjoyment compared to the more common differential incentives approach. Testing these strategies on a common ground will add to previous research on adaptive designs, which has been inconclusive about which approach works best for stimulating respondents´ participation and engagement. Furthermore, the treatments will not only be compared and evaluated with respect to their effects on participation, but also by being mindful about other, potential unintended, consequences on data quality in the long run. The project will be implemented within a mixed-mode panel study, the GESIS Panel, and thus accounts for the increasing importance of self-administered mixed-mode studies, which have been on the rise not only since the COVID-19 pandemic due to their high benefit-cost ratio. In addition, the transferability of the developed methodology to other panel studies will be investigated. This includes building similar machine learning models for predicting nonresponse in the Family Research and Demographic Analysis (FReDA) panel and the German Socio-Economic Panel (GSOEP) to assess the design-dependency of the prediction results and thus the potentials of implementing prediction-based adaptive designs in other infrastructures.

• DFG Programme: Research Grants