Each week, we publish layman’s abstracts of new articles from our prestigious portfolio of journals in statistics. The aim is to highlight the latest research to a broader audience in an accessible format.The article featured today is from the Canadian Journal of Statistics with the full article now available to read here.
Ozturk, O., Kravchuk, O. and Brown, J. (2022), Two-stage cluster samples with judgment post-stratification. Can J Statistics. https://doi.org/10.1002/cjs.11644
In surveys, members of populations are often presented in clusters – non-overlapping groups that occur naturally or are arranged to meet constraints in data collection. For this situation, a two-stage sampling design selects a sample in this way: a sample of clusters first, and then a sample of individual members in each cluster. The simplest and straightforward design, which selects simple random samples (SRS) at both stages, often has low efficiency. When an inexpensive ranking method is available for dividing clusters and individual members into ranking groups in relation to the variable of interest, the information content of the SRS at each stage can be improved. Each SRS of clusters and/or members can be judgment post stratified (JPS) into ranking groups based on their positions in small comparison samples. Those comparison samples will only consider the ranks of observations, and thus not increase the overall cost of the quantification, while allowing the sample sizes in the two-stage design to be reduced. This paper considers the efficiency of the two-stage cluster design when JPS sampling is incorporated in each or in only one of the stages. The results are of interest in applications when a two-stage cluster design would be a good choice and when there is additional information available for clusters and individual members that allows observations to be easily ranked in relation to the property for which the population mean or total is being estimated.