Defining the sample designs for small area estimation
share
The paper delineates a proper statistical setting for defining the sampling design for a small area estimation problem. This problem is often treated only via indirect estimation using the values of the variable of interest also from different areas or different times, thus increasing the sample size. However, let us suppose the domain indicator variables are available for each unit in the population. In that case, we can define a sampling design that fixes the small-area samples sizes, thereby significantly increasing the accuracy of the small area estimates. We can apply this sampling strategy to business surveys, where the domain indicator variables are available in the business register and even household surveys, where we can have the domain indicator variables for the geographical domains.
READ FULL TEXT