The population for LSMS consists of Republic of Serbia residents, excluding Kosovo and Metohija . The sampling frame for the LSMS was based on the Enumeration District (ED) delineated for the 2002 Serbia Census, excluding those with less than 20 households. It is estimated that the households in the excluded EDs only represent about 1 percent of the population of Serbia.
The sampling frame also excludes the population living in group quarters, institutions and temporary housing units, as well as the homeless population: these groups also represent less than 1 percent of the population, so the sampling frame should cover at least 98 percent of the Serbian population.
Stratification was done in the same way as for the previous LSMSs. Enumeration District werestratified according to:
- Region in 6 strata (Vojvodina, Belgrade, West Serbia, Sumadija and Pomoravlj e, East Serbia and South East Serbia).
- Type of settlement (urban and other).
The allocation of EDs according to region and type of settlement was propoI1ionai to the number of occupied dwellings, adjusted to provide sufficient precision of estimates at the regional level.
To provide optimal sample sizes in each region we decided that the minimum number of allocated EDs to each stratum should be 60. The result of this procedure was a slight deviation from strictly proportional allocation.
The sample size for LSMS 2007 was 71 40 households from 510 selected EDs. Within each ED 14 occupied dwellings were selected. From each selected occupied dwellings one household was selected (using a Kish Grid). The sample size was determined according with the aim of achieving 5,000 household interviews with an expected non-response rate of around 30%. The final response rate was 78%, producing a sample size of 5,557 households.
