Sampling Procedure
(i) Sampling frame
The sampling frame used is a list of clusters or primary sampling units (PSUs) with their respective numbers of private households. The master sample of PSUs was drawn in 2005, based on the Relative Development Index (RDI) derived from the population census data. This index is used as the spatial stratification factor and is based on 12 variables encompassing housing and living conditions, literacy and education, and employment derived from the 2000 Housing and Population Census to rank PSUs. It will be updated at the next population census. The sampling frame as of Census 2000 consisted of
298,655 households (290,004 in Mauritius and 8,561 in Rodrigues).
(ii) Sampling method
The sampling method is a stratified two-stage sampling design. At the first stage, PSUs are selected from the PSU master sample with probability proportional to size. At the second stage of the sampling process, a fixed number of households (10) are selected by systematic sampling from each selected PSU with religion, household size and the average monthly expenditure of the household as stratification criteria. In 2010, 11,280 households were surveyed at the rate of 940 per month and the response rate for 2010 was 92.1%. It should be noted that the households that could not be surveyed were replaced by households with similar characteristics.
Weighting
Since the sampling design used is not self-weighting, individuals' records have to be weighed in order to make the weighted records represent the population of inference as closely as possible. This is done in two stages: the first one is to compensate for unequal selection probabilities and the second one is to make the resultant weighted population estimates from the sample conform to independent population estimates.
(i) Quarterly estimates
Quarterly estimates are based on a sample of 2,820 households. The first set of weights (PSUWGT) consists of sampling weights based on the number of households in selected PSUs as per listing carried out in 2008. Estimates of the total number of persons by age group and sex worked out from the survey data after weighting are compared with independent population estimates (Demography figures). Correction factors are then applied to the PSUWGT for consistency.
(ii) Annual estimates
The sample data for the four quarters are merged and from the 11,280 households, the same weighting system is applied in two stages. However, in the first stage, the household weight is based on 20 households instead of 10 as a PSU is rotated twice during a year.
