Sampling, Weighting, and Imputation

The project “Sampling, Weighting, and Imputation” tackles the survey-methodological problems that arise during the preparation and analysis of panel studies conducted by the Leibnitz Institute for Educational Trajectories. Therefore, one of the core tasks of this project is the preparation of sampling designs for the establishment of (representative) samples and for possible disproportionate oversampling of subgroups that are of special interest, such as specific groups of migrants or people with unconventional educational trajectories. In this context, design and extrapolation weights are calculated, supplied, and documented. Moreover, the accurate analysis of longitudinal survey data requires the consideration of default processes due to nonresponse or refusals. Consequently, another essential task of the project is modeling and analyzing such processes as well as describing the selectivity that evolves in the course of longitudinal studies. The selectivity that is demonstrated is used for the calculation of survey weights. When survey data are analyzed, these weights compensate for nonresponse and refusals and enable the deduction of routines for the imputation of missing values. If available, data from the official statistics are consulted to improve data quality.


Providing statistical advice on data analysis, particularly concerning panel analyses as well as weighting and imputation, in consideration of complex patterns of incomplete data is another important aspect of the work carried out by the project team.  


Moreover, the project is highly involved in the assurance of survey quality by reporting on panel progression and by examining reports of survey institutes on field work.

 

Contact: statistik(at)lifbi.de