LAUSR

This research proposes a decision framework that allows for the identification of the most suitable methods to construct stable composite indicators that capture the concept of multidimensional social phenomena. This decision framework is applied to discover which method among six best represents the social exclusion of eight medium-sized Brazilian cities. The results indicate that space is important in the definition and performance of the method, and ease methods to apply present the best performance. However, one of them fails to capture the concept of the multidimensional phenomenon in two cities. The research makes six important contributions to the literature. First, it offers a decision framework for choosing the best-fit method to construct a composite social indicator. Second, it shows to what extent geographic space matters in defining the best-fit method. Third, it identifies the best-fit method regarding stability and linkage with the conceptually most significant indicator of social exclusion. Fourth, it reveals the methods to be avoided, given their poor performance. Fifth, it indicates the mathematical properties that best represent composite social phenomena. Sixth, it illuminates the debate on social exclusion from a geographical and public policy perspective.