Project Details
Projekt
Explaining nonresponse and countering nonresponse bias in self-administered panel surveys
Subject Area
Empirical Social Research
Term
since 2024
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 530859036
Nonresponse is a major challenge for the quality of survey data. Especially in panel surveys, the decreasing number of participants over the course of the panel is a major problem. The planned project consists of two parts: "Understanding Nonresponse" and "Countering Nonresponse Bias". The first part will examine the extent to which characteristics of the participants and the survey influence participation in the follow-up wave in the GESIS Panel. In particular, it is analysed whether different characteristics explain nonresponse at different stages of the panel, whether the direction or strength of the effects change and whether the characteristics depend on each other. It will also be examined to what extent the results from the GESIS panel can be transferred to comparable panel studies. Due to the large number of (potentially correlated) characteristics, machine learning methods are used for the analysis. The use of methods of so-called double machine learning enables the unbiased estimation of causal relationships. These are estimated to be dynamic and heterogeneous. The methods of double machine learning will be (further) developed for the analysis of panel data within the project. The newly developed methods will be published in the form of R and Python packages. In the second part of the project, it will be investigated to what extent different interventions (additional monetary incentive, sending of SMS or emails including a summary of the study results between the waves) contribute to the change in response behaviour. A targeted design will be developed to attribute the interventions to panel members in such a way that differences in participation behaviour between them are reduced as much as possible.
DFG Programme
Research Grants
