Explaining nonresponse and countering nonresponse bias in self-administered panel surveys
Panel surveys are an important tool for investigating people's opinions, attitudes and preferences. They are irreplaceable for measuring social change. The validity and generalizability of panel surveys depends crucially on the panel participants being a good representation of the population.
Probability-based surveys ensure that every person in the target population has a (known) chance of being included in the panel by drawing them randomly from available sampling frames, such as official registers like the residents' registration office in Germany. The people included in the sample are then invited to participate in the panel study. In order to be representative of the target population, the participants in a panel should be and remain a random group of the people included in the sample.
Studies show that response rates to surveys are falling. For various reasons, some respondents drop out of the survey. This is a major problem with panel surveys in particular, as dwindling response rates accumulate across the panel waves. This can result in considerable limitations for the informative value of the survey.
The aim of the project is to contribute to the explanation of non-response in panel surveys and to develop methods to determine the bias of estimators due to non-response.
Further Project Participants
Barbara Felderer vom GESIS - Leibniz-Institut für Sozialwissenschaften