Multi stage sampling in SMART Surveys

Dear Anonymous_A_W_40:

There is no set standard procedure for carrying out sampling in SMART surveys. The SMART manual describes different sampling schemes, but the sampling method to be used for each survey must be matched to the available data on the population to be surveyed. Some populations may have a complete list of all the basic sampling units, in which case simple random sampling can be used. This was true in a survey of adolescents in Bhutanese refugee camps in Nepal. Other populations may have basic sampling units arranged in some regular pattern, so that systematic random sampling can be used. This was true in refugee camps in Macedonia where the tents were arranged in very clear rows. Both of these Sampling methods are preferable to cluster sampling because of the relatively higher precision obtained.

In populations where cluster sampling is required because of the lack of either a complete list or the regular arrangement of basic sampling units, the selection of which geographic units to use as the primary sampling unit depends entirely on what data are available and how the population is organized. I would strongly recommend reading the entire SMART manual and consulting experts before embarking on survey planning.