Stratification, Sampling and Estimation: Finding the best design for the Swedish Investment Survey

Abstract: In this thesis, different methods are used to investigate the best design for the Swedish Investment Survey. The methods used today are stratify on number of employees, Neyman-allocation, STSRS-sampling and HT-estimation. The alternative methods in this thesis are stratify on turnover, -sampling and GREG-estimation. These methods are combined into eight combinations and tested on the enterprises and their investments in the Swedish energy domains. The conclusion is that changing the design to stratify on turnover, -sampling and GREG-estimation will provide a smaller standard deviation in most of the energy domains.