Multidimensional Sampling of Farms within R: A Successful Kazakh-German Cooperation

Sven Schmiedel, PhD, (
Federal Statistical Office of Germany
Meyram Seydazim, (
Committee on Statistics of Kazakhstan
Zhandos Kozbanov, (
Committee on Statistics of Kazakhstan


Within the project “strengthening the national statistical system of Kazakhstan”, the Federal Statistical Office of Germany (DESTATIS) and the Committee on Statistics in Kazakhstan (CSK) intensively collaborated in the field of sampling methodology and the related technical implementation. One part of the project was to implement the multidimensional sampling methodology for agricultural surveys applied by the CSK in the statistical software “R” and thereby automatize the work process at CSK.
The frame of the sample is the agricultural register. Different types of crops are sampled together in a multidimensional approach. Key variable for the sampling design is the seed area for each crop, which is used to calculate the inclusion probability. Hence, the sample is proportional to size. The inclusion probability is used to select farms randomly. The method includes additionally an exponent, which reduces the probability of very large farms to be included.
The method was implemented successfully in “R” and thereby reduced massively the work load of the personnel of the division of sampling surveys in Kazakhstan.

Keywords: Sampling, Agriculture, International Cooperation, Asia, Kazakhstan, Europe, Germany, Statistical Software R
JEL Classification: C83, Q12

[Full Text]

Romanian Statistical Review 2/2016