What does it mean that the sample followed a two-stage, replicated and rotable design?
Nigeria’s General Household Survey is an example of a two-stage, replicated, and rotating design. For the 2006 survey, each state was demarcated into 120 Enumeration Areas and then 12 replicates were independently drawn from these EAs (i.e. each replicate included 10 EAs). Each replicate was drawn so as to be representative of the state population. The GHS employs a rotating sample design, which means that part of the sample overlaps in consecutive years. This was done by rotating out half of the replicates each year.
For the first stage of sampling, 60 of the 120 EAs were randomly chosen within each state and half of the replicates from these 60 EAs were then used in the second stage of sampling. These chosen Enumeration Areas are the Primary Sampling Units (PSU). For the second stage of the sample selection, 10 Housing Units were systematically selected in each of the 60 PSU from a random starting point on a list of all replicate households. Thus, each state had a sample of 600 households chosen for interview.
Hope this helps.
For the first stage of sampling, you say that 60 EAs or 6 replicates were randomly selected within each state. Then, half of these 60 EAs (i.e. 30 EAs) were used in the second stage of sampling when 10 household were chosen within each EA. But this equals 300 households chosen for interview within state.
By second stage sampling you mean selecting secondary sampling units (households), right?
What am I missing?
Keep in mind that each Enumeration Area has 12 independently-drawn replicate samples. In the first stage of sampling, 60 EAs are chosen out of a total of 120 EAs. In the second stage, 10 households (i.e. secondary sampling units) are chosen from each of these 60 EAs, but the households eligible for this selection are limited to just 6 of the 12 replicates within an EA. Thus, the final sample to be interviewed is 600 households (= 60 EAs * 10 households).
I hope this helps to clarify.