Marco Ballin (ballin@istat.it)
Giulio Barcaroli (barcarol@istat.it),
Elena Catanese (catanese@istat.it),
Marcello D’Orazio (madorazi@istat.it),
Italian National Institute of Statistics (Istat)
Abstract
Usually sample surveys on enterprises and farms adopt a one stage stratified sampling design. In practice the sampling frame is divided in non-overlapping strata and simple random sampling is carried out independently in each stratum. Stratification allows for reduction of the sampling error and permits to derive accurate estimates. Stratified sampling requires a number of decisions strictly related: (i) how to stratify the population and how many strata to consider; (ii) the size of the whole sample and corresponding partitioning among the strata (so called allocation). This paper will deal mainly with the problem (i) and will show how to tackle it in the R environment using packages already available on the CRAN.
Keywords: stratified random sampling; multipurpose surveys, optimisation
JEL Classification: C83