An up to date problem in household and individual surveys is the nonparticipation of units which may lead to significant nonresponse biases. One approach to overcome the nonparticipation in these surveys is the use of access panels. Though leading to expected higher response rates, the self-selection process at the recruitment stage urges the needs of a correction of the estimates in access panels. This can be done directly either by calibrating the estimates to the population of interest or by additionally using response propensity scores. The German Microcensus serves as recruitment pool for the access panel of German Official Statistics that is called Dauerstichprobe (DSP) with its voluntary nature of participation. The generation of a sample from the DSP must be treated as a consequence of multiple sampling stages with different inclusion probabilities at each stage. Therefore the whole selection process has to be taken into account when computing estimates based on surveys drawn from the DSP, e.g. the German Subsample of the European Union Statistics of Income and Living Conditions (EU-SILC). Considering estimates for change in time adds further complexity, especially when being interested in poverty estimates. The paper is built on earlier findings of ENDERLE, MÜNNICH AND BRUCH (2011) and AMAROV AND RENDTEL (2011). It has the aim to compute weights of all selection stages and examine their influence on measures of level and change over time. The performance of the weighting is measured in a close-to-reality simulation study. The generation of the survey samples is done by the application of a three stage selection experiment. That is, each replication of the simulation is a complete randomization of the sampling stages Microcensus (MC), access panel (DSP) and survey of interest (e.g. D-SILC) when drawing from the sampling frame (i.e. synthetic data of the German population).