The finite population bootstrap method is used as a computer-intensive alternative to estimate the sampling distribution of a sample statis-tic. The generation of a so-called “bootstrap population” is the necessarystep between the original sample drawn and the resamples needed to mimicthis distribution. The most important question for researchers to answer ishow to create an adequate bootstrap population, which may serve as a close-to-reality basis for the resampling process. In this paper, a review of someapproaches to answer this fundamental question is presented. Moreover, anapproach based on the idea behind the Horvitz-Thompson estimator allow-ing not only whole units in the bootstrap population but also parts of wholeunits is proposed. In a simulation study, this method is compared with a moreheuristic technique from the bootstrap literature.

The motivation behind considering the use of indirect questioning designs is their possible positive effect on the respondents' willingness to cooperate. Whereas the privacy protection objectively offered by these methods has a direct effect on the estimator's efficiency, it is the subjectively perceived protection which affects the respondents' willingness to cooperate. For the discussion of these different aspects of privacy protection, a family of randomized response techniques enabling the tailoring of the design's privacy protection to the respondents is presented as representative of indirect questioning designs. Measures are suggested that formalize how the objectively offered and subjectively perceived privacy protection may differ. Different features of randomized response questioning designs, influencing the perceived privacy protection, are discussed particularly for the "crosswise model" in order to avoid underestimations of the true levels of privacy protection, which would be counter-productive with regard to the respondents' cooperation propensity.

The subject of the EU-project DACSEIS is data quality in complex surveys. In a simulation study samples have been selected out of a pseudopopulation, which was generated from the data of one Austrian Microcensus (AMC). This selection was done according to the most important facts of the sampling design of the AMC until 2003. Nonresponse of different amount was added to these samples according to a prede ned mechanism.The simulation results show, that the AMC-estimator (without iterative proportional tting) for a number of units, given the sampling method used, is less ef cient than it would be with simple random sampling and that it is possible to give a direct variance estimation for it, when there is full response. Weighting for nonresponse produces biased estimators, for which it is still possible to have a good direct variance estimation. But the bias of the AMC-estimator decreases, when appropriate auxiliary information for different imputation techniques (Single and Multiple Imputation) is used. The best estimation of the variance of such imputed estimators within the used variance estimation techniques are appropriate bootstrap techniques for single imputation and the implied variance estimation for multiple imputation.

No abstract