Access panels (AP) consist of persons who deliberately participate in surveys. The advantages of sampling from an AP are - among others - a simpler sampling of sub-populations for surveys on special topics, faster fieldwork with the opportunity to implement web surveys. These desirable features are obscured by an important drawback, however. The percentage of persons who can be recruited for an AP is as low as about 10 percent. This raises the question whether inference on the basis of an AP is flawed by the self-selection process underlying the formation of the sampling basis. If this is indeed the case, appropriate tools for data collected in an AP are required.
This project wants to answer this question by using the example of the "Dauerstichprobe", an AP recruited from the German Microcensus. Our approach uses a sequential sampling scheme, with selection probabilities on the different stages that have to be estimated. A simulation study will be used to validate the results. The project will study the impact of different recruiting strategies and interview modes on the participation propensities.