Illustration of the construction of a multidimensional index with the Urban Vulnerability index

The analysis techniques used in the construction of urban vulnerability indices are dominated by dimension reduction strategies, among which principal component analysis (PCA) stands out. Apart from its virtues, such as simplicity and interpretability, PCA has certain limitations that must be borne in mind. On the one hand, PCA, like factor analysis (FA), treats all variables together, under the assumption that they are all related to the underlying dimensions, so that they do not allow for an organised analysis along dimensions. This makes comparability between annuities very difficult, due to the fact that there is no certainty that the information contained in the first factor of the analysis of one annuity corresponds to that of the following annuity. To overcome these limitations, structural equation models are proposed in the literature, which make it possible to configure a much more stable factor structure by controlling the various dimensions that group together the indicators involved in the analysis, making them comparable over time. The session will illustrate the methodology followed to construct the Urban Vulnerability Index following this procedure, and the main results obtained will be presented.