This paper describes a method of sampling that has been developed for use on national household surveys in Bosnia and Herzegovina (BiH). The design and implementation of household surveys in BiH is particularly challenging for a number of reasons, including the absence of population registers or other sampling frames, the absence of relevant Census data or other population statistics, and continuing rapid change in the population of occupied dwellings (building, renovation, and associated population movement). The paper sets out the components of a sampling method that is both practical and statistically efficient. Both the practical and statistical arguments for each proposed component
are presented. The results of a field test of the method are summarised. The sample design adopted
for the BiH Household Budget Survey 2003-04, the first survey to use the new methodology, is
described and the paper also sets out how the design can be adapted to the needs of other surveys.
The sampling solutions adopted and described in this paper are likely to be applicable also to other
situations in which analogous challenges exist