Avariable clustering approach to quality of life in local territories -the Clust Of Var method

Analyzing and measuring well-being of citizens is a major issue for public policies. It is now widely accepted that this implies overcoming monetary and economic issues to encompass the multiple dimensions defining well-being: health, education, natural environment, social ties, or participation in civic life for instance. Recently, it has acquired unprecedented importance with regard to the multiplication of the work of Observatories, Statistical Institutes and National and International Organizations. In France, this has resulted in proposals of regional human development indices and territorial indicators of quality of life. This paper contributes to this literature. It uses a statistical approach based on variable clustering for the analysis and measurement of quality of life at the scale of local territories. The features of the ClustOfVar method -in particular the simultaneous construction of clusters of variables and the definition of associated synthetic variables -make it possible to respond to the double challenge of reducing the size of data and revealing the multidimensionality of living conditions. This work uses the dataset compiled by the French National Institute of Statistics and Economic Studies (Insee) for the analysis of quality of life in France. The results highlight the associations between variables and reveal the empirical components that structure living conditions. The mapping of synthetic variables and the calculations of spatial autocorrelation indices confirm the existence of spatial interactions operating at different scales.