Sampling
Deviations from Sample Design
The sample design was done by the NSO Sampling and Analysis Division. A multi-stage sampling strategy was used and the initial target size of the sample was 7,145 households. Considering the probability of non-response due the sensitive nature of the survey, the sample size was increased to 7,860 households, assuming a non-response rate of 10 per cent. The sample size was designed to produce reliable estimates at national level and with margins of error less than 10 per cent at urban/rural level, and less than 20 per cent at province level. The sampling frame includes units from all 21 provinces and nine districts of the capital city. Khesegs (parts of sub-districts in the capital city) and baghs (the smallest administrative unit in provinces) were selected as Primary Sampling Units (PSUs). Data was collected from a total of 524 PSUs covering 104 khesegs and 420 baghs, and 15 households were selected from each PSUs. The sample design process was completed as follows:
