Department of statistical coordination and research, Institut national de la statistique et de la démographie, Ouagadougou, Burkina Faso
Probabilistic sampling is theoretical the best method to select representative sample from target population. However, this result does not always hold in practice because some selected sampling units are missed during survey. The aim of this paper is to categorize the 45 provinces of Burkina Faso in terms of household surveys complexity in order to adapt specific methodology to each group of similar provinces.
We applied hierarchical clustering based on principal components analysis on provinces data and found five clusters. According to these results, two provinces (Kouritenga and Loroum) were the most hard-to-survey. Therefore, adapted methodology should be applied to each group of provinces during survey implementation to maximize data quality.
Keywords: hard-to-survey, hierarchical clustering, FactoMineR.
Pages: 104 – 109 | Full PDF Paper